HARL enables agents to execute cooperative tasks by adopting agent-specific policies. Most of existing HARL methods use individual policy neural networks to ensure monotonic improvement, which leads to substantial computational overhead. The proposed SDE-HARL overcomes this limitation by decomposing each agent's policy neural network into a lightweight local neural network and a global neural network executed at an edge server. Each local neural network generates and sends a compressed latent representation to the edge server, which aggregates the representations and produces agent-specific inferences. As such, SDE-HARL allows to significantly save computing and networking resources while preserving agent-specific behavior. A key feature of SDE-HARL is grouping agents with similar roles via a role-aware mechanism and share partial parameters in their global networks, while an identity-aware mechanism is introduced to promote behavioral diversity among agents within the same group. We prototyped SDE-HARL on an experimental testbed composed of a Jetson Nano and Raspberry PI to measure latency and network resource consumption. We evaluated SDE-HARL's performance on several benchmark datasets, including Google Research Football and StarCraft II. Experimental results show that SDE-HARL reaches up to 90% win rate while reducing latency, energy consumption, and networking overhead respectively by 2x, 2.5x, and 5x compared to existing work.

The exponential growth of Wi-Fi-enabled devices and the changing nature of online activities highlight the urgent need to find innovative solutions to solve the ever-growing spectrum crunch. To address this challenge, the IEEE 802.11ax standard breaks away from traditional approaches leveraged in previous standards by jointly enabling orthogonal frequency-division multiple-access (OFDMA) and multi-user MIMO (MU-MIMO). While IEEE standardization groups are working to increase the network capacity (IEEE 802.11be) and its reliability (IEEE 802.11bn), there are still serious security issues in Wi-Fi. Some recent work has revealed that MU-MIMO transmissions can be thwarted by a malicious user that interferes with the procedure followed to set up simultaneous transmissions to multiple stations (STAs). In this work, we show that a similar attack is also effective in OFDMA MU-MIMO transmissions. Specifically, a malicious user in the network can alter the precoding procedure by transmitting adversarial feedback during the channel sounding phase, thus increasing the bit error rate (BER) experienced by legitimate STAs to up to 0.5 depending on the portion of feedback that is poisoned. We shared the code to implement our attack for reproducibility purposes and to ease its integration into digital twin frameworks of Wi-Fi networks to study and evaluate effective countermeasures.

With the growing demand for device-free and privacy-preserving sensing solutions, Wi-Fi sensing has emerged as a promising approach for human pose estimation (HPE). However, existing methods often process vast amounts of channel state information (CSI) data directly, ultimately straining networking resources. This paper introduces TinySense, an efficient compression framework that enhances the scalability of Wi-Fi-based human sensing. Our approach is based on a new vector quantization-based generative adversarial network (VQ-GAN). Specifically, by leveraging a VQGAN-learned codebook, TinySense significantly reduces CSI data while maintaining the accuracy required for reliable HPE. To optimize compression, we employ the K-means algorithm to dynamically adjust compression bitrates to cluster a large-scale pre-trained codebook into smaller subsets. Furthermore, a Transformer model is incorporated to mitigate bitrate loss, enhancing robustness in unreliable networking conditions. We prototype TinySense on an experimental testbed using Jetson Nano and Raspberry Pi to measure latency and network resource use. Extensive results demonstrate that TinySense significantly outperforms state-of-the-art compression schemes, achieving up to 1.5 × higher HPE accuracy score (PCK20) under the same compression rate. It also reduces latency and networking overhead, respectively, by up to 5× and 2.5×. The code repository is available online at https://github.com/icclabo/CloudSense.

The rise of Open RAN is reshaping the cellular landscape, enabling unprecedented NextG applications. Unfortunately, Open RAN networks are riddled with vulnerabilities that can be leveraged to degrade UE performance. Interference that disrupts a designated target UE are among the most harmful due to their subtlety and they have been shown to completely eliminate a target UE’s uplink data throughput. Despite the success of existing attacks, some important problems remain; (1) the interferer cannot practically determine the success of its attack since it does not have access to the target UE’s throughput and (2) completely eliminating the target UE’s uplink throughput (as does the current literature) may be unnecessary and is likely to result in the interferer’s detection. In this work we formulate a subtle interference technique that utilizes feedback from the target UE (or set of target UEs) to estimate target UE throughput and tune the level of interference dynamically. This allows the interferer to degrade the target UE throughput to a desired level. Furthermore, we expand the attack surface from a single target UE to a set of target UEs, and present two unique methods of generating the interference signal, making our approach more complex than existing work. We implement a prototype of our approach using OpenAirInterface and validate its performance using Northeastern University’s Colosseum testbed as well as simulations and find that we can degrade the throughputs of a set of target UEs 39.1% closer on average to the desired target throughputs than a baseline. As such, our interference approach highlights the need for increased security measures in Open RAN networks.

As the Open Radio Access Network (O-RAN) paradigm is transforming thecellular landscape, third-party Extensible Applications (xApps) deployed inthe Near-Real-Time RAN Intelligent Controller will play a critical role infacilitating next-generation cellular networks. Such networks will rely heav-ily upon robust data-driven Artificial Intelligence and Machine Learning(AI/ML) solutions that can perform fine-grained RAN control at the mil-lisecond timescale. As xApps represent the point of contact between theseadvanced AI/ML solutions and the RAN itself, understanding these impor-tant O-RAN applications is essential for researchers in the wireless domainand will be critical to the full-scale integration of AI/ML into the next gener-ation of cellular networks. Motivated by a lack of relevant and up-to-date sur-veys on this topic, we provide a comprehensive overview of xApps, includingimportant background information, current xApp development techniques, adiscussion of xApps and AI/ML, state-of-the-art use cases, vulnerabilities inthe xApp architecture, and other critical research challenges. We summarizethe plethora of available information on this topic by highlighting the keyissues and presenting a cohesive message that provides the necessary back-ground to the reader and isolates the critical research issues that currentlyremain unaddressed. As such, this survey is a one-stop-shop for researchersseeking a solid grasp of the state-of-the-art and a clear outline of the centralresearch challenges relevant to xApps in cellular networks.

There has been significant progress in Wi-Fi sensing applications pertaining to home surveillance, remote healthcare, and home entertainment among others. However, most of the work leverages manual extraction of Channel State Information (CSI) from Wi-Fi network interface card (NIC) and targets single-subject sensing. In this work, we devise a simultaneous multi-subject sensing strategy that can adapt to different environments and people being monitored. Si-FI leverages standard-compliant beamforming feedback information (BFI) as a proxy of CSI to characterize the propagation environment. Unlike CSI, BFI (i) can be captured without any firmware modifications and (ii) captures the multiple channels between the access point and the stations without any direct access to the sensing devices. Thus, conversely, from existing work, the edge server in Si-FI records the BFI of the channels between Access Point (AP) and all the stations (STAs) (sensing devices) with a single capture, reducing the channel occupation, transmission and system latency dramatically. To achieve generalization over unseen environments and people, we develop a few-shot learning algorithm named Si-FI FREL to operate with beamforming feedback angles (BFAs) (compressed BFI). We validate Si-FI through an extensive data collection campaign in 3 different environments and 3 subjects performing 20 different activities simultaneously. We demonstrate that Si-FI achieves classification accuracy of up to 99 %, while Si-FI FREL improves the accuracy up to 27 % when compared to the state-of-the-art domain adaptation algorithm. Si-FI reduces the system latency by 50 % and channel occupation by 110 KB per sample for each sensing device compared to the state-of-the-art simultaneous multi-subject sensing work.

State-of-the-art covert routing in heterogeneous networks (HetNets) focuses on balancing covertness and throughput, but often overlooks explicit energy optimization. While covert communication inherently limits transmit power, meeting throughput demands without coordinated design can still lead to high energy consumption. To this end, we propose DECOR, Decentralized Energy-efficient COvert Routing framework that jointly optimizes covertness, throughput, and energy efficiency. Unlike traditional methods that use a single wireless technology, DECOR leverages the diversity of available wireless communication technologies in HetNet to enable simultaneous multi-modal routing. The core idea behind DECOR is that optimal simultaneous utilization of multiple modalities improves throughput and overall energy efficiency. It minimizes the end-to-end energy consumption while satisfying stringent constraints on throughput and covertness through two core steps: (1)link-level optimizationusing sequential least squares programming (SLSQP), and (2)network-level optimizationthrough a custom cluster-based routing strategy. DECOR introduces a novel clustering-based strategy that aggregates intra-cluster link information and delegates routing decisions to cluster heads, significantly reducing control overhead and enabling scalable, energy-efficient covert communication. Extensive numerical analysis demonstrates that DECOR significantly outperforms existing approaches in terms of energy-efficiency and data overhead.

Internet of Things (IoT) devices need to continuously adapt their wireless receiver based on the varying communication conditions to achieve the right trade-off between system-level performance and energy efficiency. Even though the environmentadaptability functions are developed in the back end of receiver chains, existing front-ends are designed for a fixed communication standard. The only adaptive wireless systems equipped with tunable front-ends use traditional optimization methods to reconfigure their circuit. The first issue is that these methods require a significant time to converge to the optimal configuration. The second limitation is that they cannot generalize to unseen propagation environments and operational conditions. In this paper, we propose Front-End Adaptation with Reinforcement Learning (FEARL) to dynamically optimize front-end circuits considering the ongoing distortion and interference levels with the aim to achieve optimal wireless system-level performance. FEARL characterizes the wireless link quality using the received baseband samples and reconfigures the front-end in real-time to realize an end-to-end optimized energy-efficient receiver. We developed a framework with the circuit simulator-in-the-loop, which is further utilized to train and evaluate the FEARL policy. The results show that FEARL is$7.8 x$times faster in responding to variations in the wireless environment, and is able to find a satisfactory circuit configuration even in unseen conditions.

Despite the success of Test-time adaptation (TTA), recent work has shown that adding relatively small adversarial perturbations to a limited number of samples leads to significant performance degradation. Therefore, it is crucial to rigorously evaluate existing TTA algorithms against relevant threats and implement appropriate security countermeasures. Importantly, existing threat models assume test-time samples will be labeled, which is impractical in real-world scenarios. To address this gap, we propose a new attack algorithm that does not rely on access to labeled test samples, thus providing a concrete way to assess the security vulnerabilities of TTA algorithms. Our attack design is grounded in theoretical foundations and can generate strong attacks against different state of the art TTA methods. In addition, we show that existing defense mechanisms are almost ineffective, which emphasizes the need for further research on TTA security. Through extensive experiments on CIFAR10-C, CIFAR100-C, and ImageNet-C, we demonstrate that our proposed approach closely matches the performance of state-of-the-art attack benchmarks, even without access to labeled samples. In certain cases, our approach generates stronger attacks, e.g., more than 4% higher error rate on CIFAR10-C.

Reconfigurability and self-optimization have become essential during radio frequency integrated circuit (RFIC) design to support the growing number of devices and fast changes in the surrounding wireless spectrum. This has created the need to develop new design approaches for RFICs based on end-to-end wireless system-level performance metrics during operation in dynamically changing communication environments. This paper introduces a CMOS mixer with a wide range of digitally tunable bias current for machine learning (ML) based adaptation. The mixer is designed to become part of a self-adaptive receiver (RX) architecture that is capable of optimizing its performance in accordance to wireless channel conditions by evaluating systemlevel parameters. The proposed mixer topology has digitally tunable bias current and a programmable helper current (PHC) circuit to maintain the voltage headroom during operation with a wide tuning range. It is capable of dynamically minimizing power consumption based on performance and wireless network requirements. Post-layout simulation results show that the power consumption can be reduced up to 8x depending on momentary performance needs, while maintaining an IIP3 greater than -2.4 dBm for the entire range of operation.

While artificial intelligence (AI) is improving the performance of O-RAN, it will also expose the network to adversarial machine learning (AML) attacks. For this reason, in this paper, we are the first to investigate AML in the context of deep reinforcement learning (DRL)-based O-RAN xApps. What separates AML in O-RAN from traditional settings is the need to design and analyze adversarial attacks based on RAN-specific Key Performance Measures (KPMs) such as transmitted bit rate, downlink buffer occupancy, transmitted packets, etc. As such, we propose the AdvO-RAN framework, which includes (i) a new adversarial perturbation generator using preference-based reinforcement learning (PbRL) to learn the perturbation that most violate the user service level agreements (SLA) and (ii) a robust training module for enhancing DRL agent resilience to the attacks in (i). We experimentally evaluate AdvO-RAN on the Colosseum network emulator. Experimental results show that AdvO-RAN can enhance xApp performance by reducing SLA violations from 44% to 27% on average and reducing by 46% the latency under the most challenging attack scenario for Ultra-Reliable Low-Latency Communications (URLLC) traffic. AdvO-RAN can improve up to 75% of throughput in the victim Enhanced Mobile Broadband (eMBB) slice users during a constant bit-rate traffic scenario.

Enabling spectrum perception with deep neural networks (DNNs) directly connected to the radio front-end is of fundamental importance to realize next-generation spectrum-aware wireless systems. As such, low-latency DNN inference in reconfigurable hardware such as Field Programmable Gate Arrays (FPGAs) is a necessary precursor to enable spectrum perception in real-world wireless systems. The key issue with existing work is that it considers DNNs that have fixed weights and architecture. On the other hand, it has been shown that dynamically changing the structure and weights of the DNN at runtime can lead to improved efficiency and adaptability. This work fills the current research gap by proposing FlexRFML, the first framework to integrate dynamic DNNs in the RF-front spectrum perception loop. FlexRFML includes High-Level Synthesis (HLS)-based design as well as customized circuits to achieve dynamic hardware reconfiguration and accelerate the DNN. We have prototyped FlexRFML on a Xilinx system-on-chip (SoC) ZCU102 by considering both modulation recognition and radio fingerprinting classification problems where the DNNs are dynamically adapted based on a preliminary classification of the input. Experimental results show that FlexRFML can decrease the inference latency by up to 35.6% with respect to static DNN inference with negligible additional hardware overhead. We pledge to release the FlexRFML hardware and software code.

The performance of multiple-input, multiple-output (MIMO) systems highly depends on the precision of channel estimates provided by the mobile users. However, the current Wi-Fi standard requires an update interval of 10 ms, irrespective of the channel dynamics. This imposes a substantial overhead for the MIMO channel estimation. Recent work mainly targets different compression strategies, potentially compromising precoding accuracy and, in turn, the network performance. In stark opposition, we propose SHRINK, a framework to dynamically adapt the feedback transmission rate to the propagation environments and performance requirements. SHRINK determines whether the users should send back their channel estimates by predicting network performance through a data-driven analysis of prior and current channel estimates. We have experimentally evaluated SHRINK using off-the-shelf Wi-Fi devices in multiple environments, including an anechoic chamber, and benchmarked its performance against several state-of-the-art approaches. Experimental results show that SHRINK reduces airtime and data overhead by 81% on average compared to the IEEE 802.11 standard without impacting the precoding performance. Moreover, SHRINK outperforms state-of-the-art approaches by an average gain of 33.6% in airtime and data overhead reduction, corresponding to an increase in throughput of 24.5%.

Test Time Adaptation (TTA) has emerged as a practical solution to mitigate the performance degradation of Deep Neural Networks (DNNs) in the presence of corruption/ noise affecting inputs. Existing approaches in TTA continuously adapt the DNN, leading to excessive resource consumption and performance degradation due to accumulation of error stemming from lack of supervision. In this work, we propose Domain-Aware Real- Time Dynamic Adaptation (DARDA) to address such issues. Our key approach is to proactively learn latent representations of some corruption types, each one associated with a sub-network state tailored to correctly classify inputs affected by that corruption. After deployment, DARDA adapts the DNN to previously unseen corruptions in an unsupervised fashion by (i) estimating the latent representation of the ongoing corruption; (ii) selecting the sub-network whose associated corruption is the closest in the latent space to the ongoing corruption; and (iii) adapting DNN state, so that its representation matches the ongoing corruption. This way, DARDA is more resource-efficient and can swiftly adapt to new distributions caused by different corruptions without requiring a large variety of input data. Through experiments with two popular mobile edge devices - Raspberry Pi and NVIDIA Jetson Nano - we show that DARDA reduces energy consumption and average cache memory footprint respectively by 1.74 x and 2.64 x with respect to the state of the art, while increasing the performance by 10.4%, 5.7% and 4.4% on CIFAR-10, CIFAR-100 and TinyImagenet.

A fundamental challenge in covert routing is that meeting both covertness and throughput requirements often leads to increased transmit power, which can significantly elevate the overall energy consumption of the network. Therefore, it is important to achieve higher throughput and better energy efficiency while maintaining the required covertness. To this end, we propose a novel simultaneous multi-modal Decentralized Energy- Efficient covert Routing approach - DEER for a multi-hop heterogeneous network (HetNet). Unlike the prevailing single-modal approaches, DEER leverages the diversity of the available wireless communication technologies for simultaneous multi-modal routing. DEER aims to minimize the end-to-end total transmit power of the whole route in a decentralized fashion while maintaining the constraints on required throughput and covertness. DEER stems into two main steps: node-level optimization followed by network-level optimization using the proposed custom-tailored Dijkstra's based link state routing protocol to meet the constraints while minimizing the end-to-end total transmit power. We demonstrate by numerical analysis that DEER improves the energy efficiency by$23.5 x$and$2.9 x$times in comparison to the baseline single-modal and naive simultaneous multimodal approaches respectively.

In this paper, we present MAGIC, a novel approach to gesture recognition utilizing mmWave multiple-input multiple-output (MIMO) Channel State Information (CSI). Unlike existing mmWave gesture recognition methods that often rely on radar signals, MAGIC leverages CSI extracted from mmWave MIMO integrated sensing and communication (ISAC) systems. While advanced radar systems, such as those operating in frequency-modulated continuous wave (FMCW) mode, can achieve high frequency and spatial resolution, they typically require dedicated sensing infrastructure, which increases system complexity. In contrast, MAGIC utilizes high-granular CSI from orthogonal frequency-division multiplexing (OFDM) systems, enabling fine spatial, temporal, and frequency-domain information for robust gesture recognition. This eliminates the need for dedicated radar transceivers, simplifying the system and reducing transmission overhead. MAGIC employs a learning-based architecture, integrating a temporal convolutional network (TCN) to classify gestures by capturing long-range temporal dependencies. To address the critical challenge of domain adaptation in gesture recognition, we propose adaptive temporal embedding network (ATEN), a meta-learning framework that combines the temporal modeling capabilities of TCN with task-specific adaptation mechanisms. We evaluateMAGIC through a comprehensive data collection campaign involving two subjects performing 10 micro gestures across three different environments, with synchronized video streams providing the ground truth. The proposed system achieves a baseline accuracy of 99.24% using TCN. The system continues to perform well – achieving up to 98.82% accuracy – when adapting to new domains using ATEN, outperforming other state-of-the-art domain adaptation methods by 14% on average.

During advanced surveillance missions, Unmanned Aerial Vehicles (UAVs) usually require the execution of edge-assisted computer vision (CV) tasks. In multi-hop UAV networks, the successful transmission of these tasks to the edge is severely challenged due to severe bandwidth constraints, and the possible node failures. To address these critical challenges, we propose a novel A 2 - UAV framework that optimizes the number of correctly executed tasks at the edge. In stark contrast with existing art, we take an application-aware approach and formulate a novel Application-Aware Task Planning Problem ( A 2 - TPP ) to optimize routing, data pre-processing and target assignment for each UAV. Our formulation explicitly takes into account (i) the relationship between CV task accuracy and image compression for the classes of interest based on the available dataset, (ii) the target positions , (iii) the current energy/position of the UAVs, and (iv) the possible node failures. We demonstrate A 2 - TPP is NP-Hard and propose a polynomial-time algorithm to solve it efficiently. We extensively evaluate A 2 - UAV through simulation and real-world experiments using a testbed composed by four DJI Mavic Air 2 UAVs. Results on image classification show that A 2 - UAV attains on average around 38% more accomplished tasks w.r.t. the state of the art, with a 400% improvement in tasks-intensive scenarios. Moreover, we show that our framework is able to reconfigure the network in case of nodes failure.

The development of Open Radio Access Network (RAN) cellular systems is being propelled by the integration of Artificial Intelligence (AI) techniques. While AI can enhance network performance, it expands the attack surface of the RAN. For instance, the need for datasets to train AI algorithms and the use of open interface to retrieve data in real time paves the way to data tampering during both training and inference phases. In this work, we propose MalO-RAN, a framework to evaluate the impact of data poisoning on O-RAN intelligent applications. We focus on AI-based xApps taking control decisions via Deep Reinforcement Learning (DRL), and investigate backdoor attacks, where tampered data is added to training datasets to include a backdoor in the final model that can be used by the attacker to trigger potentially harmful or inefficient pre-defined control decisions. We leverage an extensive O-RAN dataset collected on the Colosseum network emulator and show how an attacker may tamper with the training of AI models embedded in xApps, with the goal of favoring specific tenants after the application deployment on the network. We experimentally evaluate the impact of the SleeperNets and TrojDRL attacks and show that backdoor attacks achieve up to a 0.9 attack success rate. Moreover, we demonstrate the impact of these attacks on a live O-RAN deployment implemented on Colosseum, where we instantiate the xApps poisoned with MalO-RAN on an O-RAN-compliant Near-real-time RAN Intelligent Controller (RIC). Results show that these attacks cause an average network performance degradation of 87%.

Emerging mobile applications often require the execution of computer vision (CV) tasks based on compute-and memory-intensive deep neural networks (DNNs). Although offloading CV tasks to edge servers can decrease resource consumption at the mobile devices, it poses the challenge of handling multiple concurrent tasks with limited computing and memory capacity. In stark opposition with the existing state of the art, we tackle this challenge by jointly optimizing (i) the utilization of resources at the edge, among which memory - so far widely overlooked - and the radio resources used for task offloading; (ii) which and how many offloaded tasks should be executed; and (iii) the structure of the DNNs. First, we formulate the DNN for scalable Offloading of Tasks (DOT) problem, prove that it is NP-hard, and envision a weighted-tree-based heuristic solution, named OffloaDNN, that efficiently solves the DOT problem. We evaluate OffloaDNN through extensive numerical analysis using state-of-the-art image classification ResNet-18, as well as real-world experiments on the Colosseum emulator. The numerical results show that, in small-scale scenarios, OffloaDNN matches the optimum very closely, and, in larger-scale scenarios, increases the number of admitted offloaded tasks by 26.9 % with respect to the state of the art, while saving 82.5 % memory and 77.4% per-inference computing time. The numerical results are confirmed by the real-world validation on Colosseum.

Fifth generation (5G) and beyond cellular networks are vulnerable to security threats, primarily due to the lack of integrity protection in the Radio Resource Control (RRC) layer. In order to address this problem, we propose a real-time anomaly detection framework that leverages the concept of distributed applications in 5G Open RAN networks. Specifically, we identify Physical Layer (PHY) features that can generate a reliable fingerprint, infer in a novel way the time of arrival of uplink packets lacking integrity protection, and handle cross-layer features. By identifying legitimate message sources and detecting suspicious activities through an Artificial Intelligence (AI) design, we demonstrate that Open RAN-based applications that run at the edge can be designed to provide additional security to the network. Our solution is first validated in extensive emulation environments achieving over 85% accuracy in predicting potential attacks on unseen test scenarios. We then integrate our approach into a real-world prototype with a large channel emulator to assess its real-time performance and costs. Our solution meets the low-latency real-time constraints of 2 ms, making it well-suited for real-world deployments.

Spectrum has become an extremely scarce and congested resource. As a consequence, spectrum sensing enables the coexistence of different wireless technologies in shared spectrum bands. Most existing work requires spectrograms to classify signals. Ultimately, this implies that images need to be continuously created from I/Q samples, thus creating unacceptable latency for real-time operations. In addition, spectrogram-based approaches do not achieve sufficient granularity level as they are based on object detection performed on pixels and are based on rectangular bounding boxes. For this reason, we propose a completely novel approach based on semantic spectrum segmentation, where multiple signals are simultaneously classified and localized in both time and frequency at the I/Q level. Conversely from the state-of-the-art computer vision algorithm, we add non-local blocks to combine the spatial features of signals, and thus achieve better performance. In addition, we propose a novel data generation approach where a limited set of easy-to-collect real-world wireless signals are "stitched together" to generate large-scale, wideband, and diverse datasets. Experimental results obtained on multiple testbeds (including the Arena testbed) using multiple antennas, multiple sampling frequencies, and multiple radios over the course of 3 days show that our approach classifies and localizes signals with a mean intersection over union (IOU) of 96.70% across 5 wireless protocols while performing in real-time with a latency of 2.6 ms. Moreover, we demonstrate that our approach based on non-local blocks achieves 7% more accuracy when segmenting the most challenging signals with respect to the state-of-the-art U-Net algorithm. We will release our 17 GB dataset and code.

In this paper, we present m3MIMO a mmWave fully-digital multi-user multi-input multi-output (MU-MIMO) testbed for advanced wireless research. m3MIMO operates in the 57-64 GHz frequency range and supports up to 1 GHz of bandwidth enabling large data multiplexing in the frequency domain through orthogonal frequency-division multiplexing (OFDM). The testbed features three custom-designed Zynq UltraScale+ RFSoC-based Software Defined Radios (SDRs) empowered with the Pi-Radio fully digital transceivers. Two of these SDRs support eight transmit and receive streams each (8 × 8 MIMO), while the third SDR supports up to four channels. m3MIMO supports three different communication modes: (i) point-to-point (P2P) transmissions; (ii) single-user multi-input multi-output (SU-MIMO), where multiple streams are transmitted to a single end-device; and (iii) MU-MIMO, where two devices are simultaneously served by a single transmitter. To showcase the m3MIMO's versatility, we present two research use cases: tracking-based beamforming and mmWave-based sensing. We will open-source the m3MIMO code along with the relevant use-case datasets, facilitating further analysis1.

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.

Emerging mobile virtual reality (VR) systems are required to continuously perform complex computer vision tasks needing computational power that is excessive for mobile devices. Thus, techniques based on wireless edge computing (WEC) have been recently proposed. However, existing WEC methods require the transmission and processing of a high amount of video data which may ultimately saturate the wireless link. In this paper, we propose a novel sensing-assisted edge computing (ISAC-EC) approach to address this issue. ISAC-EC leverages knowledge about the physical environment to reduce the end-to-end latency and overall computational burden by transmitting to the edge server only the relevant data for the delivery of the service. Our intuition is that the transmission of the portion of the video frames where there are no changes with respect to the previous frames can be avoided. Through wireless sensing, only the part of the frames where any environmental change is detected is transmitted and processed. We evaluated ISAC-EC by using a 10K 360°camera with a Wi-Fi 6 sensing system operating at 160 MHz and performing localization and tracking. Experimental results show that ISAC-EC reduces both the channel occupation and end-to-end latency by more than 90% while improving the instance segmentation and object detection performance with respect to state-of-the-art WEC approaches. For reproducibility purposes, we pledge to share our dataset and code repository.

The Terahertz (THz) band (0.1–10 THz) is projected to enable broadband wireless communications of the future, and many envision deep learning as a solution to improve the performance of THz communication systems and networks. However, there are few available datasets of true THz signals that could enable testing and training of deep learning algorithms for the research community. In this paper, we provide an extensive dataset of 120,000 data frames for the research community. All signals were transmitted at 165 GHz but with varying bandwidths (5 GHz, 10 GHz, and 20 GHz), modulations (4PSK, 8PSK, 16QAM, and 64QAM), and transmit amplitudes (75 mV and 600 mV), resulting in twenty-four distinct bandwidth-modulation-power combinations each with 5,000 unique captures. The signals were captured after down conversion at an intermediate frequency of 10 GHz. This dataset enables the research community to experimentally explore solutions relating to ultrabroadband deep and machine learning applications.

In this paper, we present SDR-LoRa, an open-source, full-fledged Software Defined Radio (SDR) implementation of a LoRa transceiver. First, we conduct a thorough analysis of the LoRa physical layer (PHY) functionalities, encompassing processes such as packet modulation, demodulation, and preamble detection. Then, we leverage on this analysis to create a pioneering SDR-based LoRa PHY implementation. Accordingly, we thoroughly describe all the implementation details. Moreover, we illustrate how SDR-LoRa can help boost research on the LoRa protocol by presenting three exemplary key applications that can be built on top of our implementation, namely fine-grained localization, interference cancellation, and enhanced link reliability. To validate SDR-LoRa and its applications, we test it on two different platforms: (i) a physical setup involving USRP radios and off-the-shelf commercial devices, and (ii) the Colosseum wireless channel emulator. Our experimental findings reveal that (i) SDR-LoRa performs comparably to conventional commercial LoRa systems, and (ii) all the aforementioned applications can be successfully implemented on top of SDR-LoRa with remarkable results. The complete details of the SDR-LoRa implementation code have been publicly shared online, together with a plug-and-play Colosseum container.

Cognitive communication utilizes transient openings in the spectrum to communicate opportunistically, which is a promising technique to enable more efficient spectrum usage in an increasingly congested spectrum environment. We aim to address two main challenges associated with cognitive communication: (i) spectrum sensing should be fast and energy efficient for processing a large bandwidth in a short time; (ii) the spectrum sensing approach should be able to simultaneously recognize multiple signals that are present. In this paper, we propose to address these challenges with a novel design framework that consists of a fast on-chip spectrum sensing in conjunction with a novel learning-based spectrum analysis model at the edge to enhance the optimizations for spectrum agility. We first utilize a model of a programmable analog-based high-quality factor (Q) on-chip spectrum sensor that is capable of scanning the sub-6 GHz band to detect the spectrum usage in less than 1μs. The proposed spectrum sensor also enhances the energy efficiency of the sensing. To complement the onchip spectrum sensor, a deep learning (DL) model is deployed for a fine-grained signal detection between channels in the 400 MHz to 6 GHz range, which is intended to be executed on edge devices. Simulation results show that the DL model can detect multiple different modulated signals with a mean Intersection-over-Union (IoU) of 86.8% in highly-variable bandwidth and center frequency scenarios. Finally, we present a system-level model of our framework to demonstrate the spectrum sensing and classification in the sub-6 GHz frequency band.

The next generation of mobile networks (NextG) will require careful resource management to support edge offloading of resource-intensive deep learning (DL) tasks. Current slicing frameworks treat all DL tasks equally without adjusting to their high-level objectives, resulting in sub-optimal performance. To overcome this, we proposeSEM-O-RAN, a semantic and flexible slicing framework for computer vision task offloading in NextG Open RANs. Our framework accounts for the semantic nature of object classes as well as the level of data quality to optimally tailor data compression and minimize the usage of networking and computing resources. In fact, we show that different object classes tolerate different levels of image compression while preserving detection accuracy. To address the above issues, we first present the mathematical formulation of the Semantic Flexible Edge Slicing Problem (SF-ESP), which turns out to be NP-hard. We thus define a greedy algorithm to solve it efficiently, which is also able to always select the resource allocation that yields the best resource utilization, whenever multiple allocations satisfy the DL task requirements. We evaluateSEM-O-RAN's performance through extensive numerical analysis and real-world experiments on the Colosseum testbed, considering state-of-the-art computer-vision tasks and DL models. The obtained results demonstrate thatSEM-O-RANallocates up to 169% more tasks and obtains 52% higher revenues than the state of the art.

Multi-user MIMO is a key component of modern wireless networks. As such, investigating the related security weaknesses is a compelling necessity. A major issue unveiled by existing work is that adversaries can “poison” the channel information feedback reported to the beamformer to decrease the performance experienced by a legitimate user. Prior work, however, assumes that the feedback is reported in an uncompressed fashion, which is not the case in current wireless standards such as Wi-Fi or 5G. In this work, we first show that assuming uncompressed feedback leads to overestimating the attack effectiveness by up to 60%. Next, we formulateACFP(Adversarial Compressed Feedback Problem), a novel non-convex constrained optimization problem to find the compressed feedback that maximizes a victim’s bit error rate (BER) while satisfying maximum power constraints. We proposeWHACK(Wireless Harmful Adversarial Compressed feedbacK), a new algorithm to solveACFPand find the malicious compressed feedback based on the convexity of the objective function and constraint using a nonlinear conjugate gradient method.WHACKhas been prototyped and extensively evaluated with off-the-shelf Wi-Fi devices. Experimental results show that it maximizes the victim’s BER, while modifying less than 60% of the feedback. Our dataset and code are available.

Modern IEEE 802.11 (Wi-Fi) networks extensively rely on multiple-input multiple-output (MIMO) to significantly improve throughput. To correctly beamform MIMO transmissions, the access point needs to frequently acquire a beamforming matrix (BM) from each connected station. However, the size of the matrix grows with the number of antennas and subcarriers, resulting in an increasing amount of airtime overhead and computational load at the station. Conventional approaches come with either excessive computational load or loss of beamforming precision. For this reason, we propose SplitBeam, a new framework where we train a split deep neural network (DNN) to directly output the BM given the channel state information (CSI) matrix as input. The DNN is designed with an additional “bottleneck” layer to “split” the original DNN into a head model and a tail model, respectively executed by the station and the access point. The head model generates a compressed representation of the BM, which is then used by the AP to produce the BM using the tail model. We formulate and solve a bottleneck optimization problem (BOP) to keep computation, airtime overhead, and bit error rate (BER) below application requirements. We perform extensive experimental CSI collection with off-the-shelf Wi-Fi devices in two distinct environments and compare the performance of SplitBeam with the standard IEEE 802.11 algorithm for BM feedback and the state-of-the-art DNN-based approach LB-SciFi. Our experimental results show that SplitBeam reduces the beamforming feedback size and computational complexity by respectively up to 81 % and 84 % while maintaining BER within about 10−3of existing approaches. We also implement the SplitBeam DNNs on FPGA hardware to estimate the end-to-end BM reporting delay, and show that the latter is less than 10 milliseconds in the most complex scenario, which is the target channel sounding frequency in realistic multi-user MIMO scenarios. To allow full reproducibility, we will release our code and datasets to the community.

To perform advanced surveillance, Unmanned Aerial Vehicles (UAVs) require the execution of edge-assisted computer vision (CV) tasks. In multi-hop UAV networks, the successful transmission of these tasks to the edge is severely challenged due to severe bandwidth constraints. For this reason, we propose a novel A2-UAV framework to optimize the number of correctly executed tasks at the edge. In stark contrast with existing art, we take an application-aware approach and formulate a novel Application-Aware Task Planning Problem (A2-TPP) that takes into account (i) the relationship between deep neural network (DNN) accuracy and image compression for the classes of interest based on the available dataset, (ii) the target positions, (iii) the current energy/position of the UAVs to optimize routing, data pre-processing and target assignment for each UAV. We demonstrate A2-TPP is NP-Hard and propose a polynomial-time algorithm to solve it efficiently. We extensively evaluate A2-UAV through real-world experiments with a testbed composed by four DJI Mavic Air 2 UAVs. We consider state-of-the-art image classification tasks with four different DNN models (i.e., DenseNet, ResNet152, ResNet50 and MobileNet-V2) and object detection tasks using YoloV4 trained on the ImageNet dataset. Results show that A2-UAV attains on average around 38% more accomplished tasks than the state of the art, with 400% more accomplished tasks when the number of targets increase significantly. To allow full reproducibility, we pledge to share datasets and code with the research community.

5G and beyond cellular networks (NextG) will support the continuous execution of resource-expensive edgeassisted deep learning (DL) tasks.To this end, Radio Access Network (RAN) resources will need to be carefully "sliced" to satisfy heterogeneous application requirements while minimizing RAN usage.Existing slicing frameworks treat each DL task as equal and inflexibly define the resources to assign to each task, which leads to sub-optimal performance.In this paper, we propose SEM-O-RAN, the first semantic and flexible slicing framework for NextG Open RANs.Our key intuition is that different DL classifiers can tolerate different levels of image compression, due to the semantic nature of the target classes.Therefore, compression can be semantically applied so that the networking load can be minimized.Moreover, flexibility allows SEM-O-RAN to consider multiple edge allocations leading to the same task-related performance, which significantly improves system-wide performance as more tasks can be allocated.First, we mathematically formulate the Semantic Flexible Edge Slicing Problem (SF-ESP), demonstrate that it is NP-hard, and provide an approximation algorithm to solve it efficiently.Then, we evaluate the performance of SEM-O-RAN through extensive numerical analysis with state-of-the-art multi-object detection (YOLOX) and image segmentation (BiSeNet V2), as well as realworld experiments on the Colosseum testbed.Our results show that SEM-O-RAN improves the number of allocated tasks by up to 169% with respect to the state of the art.

Thanks to the ubiquitous deployment of Wi-Fi hotspots, channel state information (CSI)-based Wi-Fi sensing can unleash game-changing applications in many fields, such as healthcare, security, and entertainment. However, despite one decade of active research on Wi-Fi sensing, most existing work only considers legacy IEEE 802.11n devices, often in particular and strictly-controlled environments. Worse yet, there is a fundamental lack of understanding of the impact on CSI-based sensing of modern Wi-Fi features, such as 160-MHz bandwidth, multiple-input multiple-output (MIMO) transmissions, and increased spectral resolution in IEEE 802.11ax (Wi-Fi 6). This work aims to shed light on the impact of Wi-Fi 6 features on the sensing performance and to create a benchmark for future research on Wi-Fi sensing. To this end, we perform an extensive CSI data collection campaign involving 3 individuals, 3 environments, and 12 activities, using Wi-Fi 6 signals. An anonymized ground truth obtained through video recording accompanies our 80-GB dataset, which contains almost two hours of CSI data from three collectors. We leverage our dataset to dissect the performance of a state-of-the-art sensing framework across different environments and individuals. Our key findings suggest that (i) MIMO transmissions and higher spectral resolution might be more beneficial than larger bandwidth for sensing applications; (ii) there is a pressing need to standardize research on Wi-Fi sensing because the path towards a truly environment-independent framework is still uncertain. To ease the experiments' replicability and address the current lack of Wi-Fi 6 CSI datasets, we release our 80-GB dataset to the community.

Cutting-edge advances in wireless networking will soon enable a new generation of safer, smarter, and more autonomous vehicles. These vehicles will rely on real-time execution of complex Deep Learning (DL) tasks as well as high-speed multimedia streaming between road users for navigation purposes. Relying entirely on cellular networks (i) puts an unnecessary burden on an already overcrowded and expensive licensed spectrum; (ii) increases the latency of edge-offloaded tasks to intolerable levels for vehicular applications. Alongside the usage of a proper network infrastructure, vehicles will need to support on-board and offloaded cooperative intelligence. On this basis, we propose Edge-V , the first framework enabling practical vehicular edge intelligence and high-speed vehicular connectivity, using only unlicensed spectrum bands. Through a DSRC link, Edge-V acquires real-time localized knowledge, and coordinates the use of point-to-point millimeter Wave (mmWave) technologies to deliver high-bandwidth connectivity between vehicles. Edge-V also foresees smart offloading if on-board computing resources are insufficient. We prototype and evaluate Edge-V in a real-world laboratory testbed, showing its advantages with respect to cellular and cloud-based approaches.

Recent advances in Wi-Fi sensing have ushered in a plethora of pervasive applications in home surveillance, remote healthcare, road safety, and home entertainment, among others.Most of the existing works are limited to the activity classification of a single human subject at a given time.Conversely, a more realistic scenario is to achieve simultaneous, multi-subject activity classification.The first key challenge in that context is that the number of classes grows exponentially with the number of subjects and activities.Moreover, it is known that Wi-Fi sensing systems struggle to adapt to new environments and subjects.To address both issues, we propose SiMWiSense, the first framework for simultaneous multi-subject activity classification based on Wi-Fi that generalizes to multiple environments and subjects.We address the scalability issue by using the Channel State Information (CSI) computed from the device positioned closest to the subject.We experimentally prove this intuition by confirming that the best accuracy is experienced when the CSI computed by the transceiver positioned closest to the subject is used for classification.To address the generalization issue, we develop a brand-new few-shot learning algorithm named Feature Reusable Embedding Learning (FREL).Through an extensive data collection campaign in 3 different environments and 3 subjects performing 20 different activities simultaneously, we demonstrate that SiMWiSense achieves classification accuracy of up to 97%, while FREL improves the accuracy by 85% in comparison to a traditional Convolutional Neural Network (CNN) and up to 20% when compared to the state-of-the-art few-shot embedding learning (FSEL), by using only 15 seconds of additional data for each class.For reproducibility purposes, we share our 1TB dataset and code repository 1 [1].

We present DeepCSI, a novel approach to Wi-Fi radio fingerprinting (RFP) which leverages standard-compliant beamforming feedback matrices to authenticate MU-MIMO Wi-Fi devices on the move. By capturing unique imperfections in off-the-shelf radio circuitry, RFP techniques can identify wireless devices directly at the physical layer, allowing low-latency low-energy cryptography-free authentication. However, existing Wi-Fi RFP techniques are based on software-defined radio (SDRs), which may ultimately prevent their widespread adoption. Moreover, it is unclear whether existing strategies can work in the presence of MU-MIMO transmitters – a key technology in modern Wi-Fi standards. Conversely from prior work, DeepCSI does not require SDR technologies and can be run on any low-cost Wi-Fi device to authenticate MU-MIMO transmitters. Our key intuition is that imperfections in the transmitter’s radio circuitry percolate onto the beamforming feedback matrix, and thus RFP can be performed without explicit channel state information (CSI) computation. DeepCSI is robust to inter-stream and inter-user interference being the beamforming feedback not affected by those phenomena. We extensively evaluate the performance of DeepCSI through a massive data collection campaign performed in the wild with off-the-shelf equipment, where 10 MU-MIMO Wi-Fi radios emit signals in different positions. Experimental results indicate that DeepCSI correctly identifies the transmitter with an accuracy of up to 98%. The identification accuracy remains above 82% when the device moves within the environment. To allow replicability and provide a performance benchmark, we pledge to share the 800 GB datasets – collected in static and, for the first time, dynamic conditions – and the code database with the community.

Today’s radio access networks (RANs) are monolithic entities which often operate statically on a given set of parameters for the entirety of their operations. To implement realistic and effective spectrum sharing policies, RANs will need to seamlessly and intelligently change their operational parameters. In stark contrast with existing paradigms, the new O-RAN architectures for 5G-and-beyond networks (NextG) separate the logic that controls the RAN from its hardware substrate, allowing unprecedented real-time fine-grained control of RAN components. In this context, we propose the Channel-Aware Reactive Mechanism (ChARM), a data-driven O-RAN-compliant framework that allows (i) sensing the spectrum to infer the presence of interference and (ii) reacting in real time by switching the distributed unit (DU) and radio unit (RU) operational parameters according to a specified spectrum access policy. ChARM is based on neural networks operating directly on unprocessed I/Q waveforms to determine the current spectrum context. ChARM does not require any modification to the existing 3GPP standards. It is designed to operate within the O-RAN specifications, and can be used in conjunction with other spectrum sharing mechanisms (e.g., LTE-U, LTE-LAA or MulteFire). We demonstrate the performance of ChARM in the context of spectrum sharing among LTE and Wi-Fi in unlicensed bands, where a controller operating over a RAN Intelligent Controller (RIC) senses the spectrum and switches cell frequency to avoid Wi-Fi. We develop a prototype of ChARM using srsRAN, and leverage the Colosseum channel emulator to collect a large-scale waveform dataset to train our neural networks with. To collect standard-compliant Wi-Fi data, we extended the Colosseum testbed using system-on-chip (SoC) boards running a modified version of the OpenWiFi architecture. Experimental results show that ChARM achieves accuracy of up to 96% on Colosseum and 85% on an over-the-air testbed, demonstrating the capacity of ChARMto exploit the considered spectrum channels.

With more than 65M people affected by epilepsy worldwide, early prediction and response to seizure onsets have become more important than ever. Cutting-edge research in implantable medical devices (IMDs) has shown that deep neural networks (DNNs) applied to intracranial electroencephalogram (iEEG) data can predict seizures up to an hour before onset. However, offloading of iEEG data to the edge/cloud is highly prohibitive, due to the sheer size of the generated data. Existing work either focuses on the DNN training phase only, or does not consider the severe energy/space limitations of IMDs. Moreover, the technical aspects of patient personalization, which allows for patient-specific hyper-parameter tuning, still remain unaddressed. In this paper, we propose a platform called AiEEG for in vivo early seizure prediction, whose DNN hardware circuitry can be reconfigured remotely without surgery. We prototype AiEEG on a system on chip (SoC) platform and demonstrate its end-to-end capabilities in seizure prediction with a population of 30 epileptic patients, with iEEG signals coming from a real-world dataset. Extensive experimental results shows that (i) our embedded and personalized DNN has an area under the curve (AUC) averaging at 0.97 and as low as zero false positives per hour (FPH) for over half the patients, an improvement of about 3.5x with respect to a non-personalized prediction method and the best for a dataset of this size when compared to the state-of-the-art; (ii) our AiEEG platform consumes 4.2x less energy than a cloud-based approach, leading to a 4x battery lifetime improvement; (iii) we are able to remotely fine-tune the DNN through partial reconfiguration as needed in about 10s.

Thanks to the ubiquitousness of Wi-Fi access points and devices, Wi-Fi sensing enables transformative applications in remote health care, home/office security, and surveillance, just to name a few. Existing work has explored the usage of machine learning on channel state information (CSI) computed from Wi-Fi packets to classify events of interest. However, most of these algorithms require a significant amount of data collection, as well as extensive computational power for additional CSI feature extraction. Moreover, the majority of these models suffer from poor accuracy when tested in a new/untrained environment. In this paper, we propose ReWiS, a novel framework for robust and environment-independent Wi-Fi sensing. The key innovation of ReWiS is to leverage few-shot learning (FSL) as the inference engine, which (i) reduces the need for extensive data collection and application-specific feature extraction; (ii) can rapidly generalize to new environments by leveraging only a few new samples. Moreover, ReWiS leverages multi-antenna, multi-receiver diversity, as well as fine-grained frequency resolution, to improve the overall robustness of the algorithms. Finally, we propose a technique based on singular value decomposition (SVD) to make the FSL input constant irrespective of the number of receive antennas. We prototype the ReWiS using off-the-shelf Wi-Fi equipment and showcase its performance by considering a compelling use case of human activity recognition. Thus, we perform an extensive data collection campaign in three different propagation environments with two human subjects. We evaluate the impact of each diversity component on the performance and compare ReWiS with an existing convolutional neural network (CNN)-based approach. Experimental results show that ReWiS improves the performance by about 40% with respect to existing single-antenna low-resolution approaches. Moreover, when compared to a CNN-based approach, ReWiS shows a 35% more accuracy and less than 10% drop in accuracy when tested in different environments, while the CNN drops by more than 45%. To allow reproducibility of our results and to address the current dearth of Wi-Fi sensing datasets, we pledge to release our 60 GB dataset and the entire code repository to the community.

Mobile devices such as drones and autonomous vehicles increasingly rely on object detection (OD) through deep neural networks (DNNs) to perform critical tasks such as navigation, target-tracking and surveillance, just to name a few. Due to their high complexity, the execution of these DNNs requires excessive time and energy. Low-complexity object tracking (OT) is thus used along with OD, where the latter is periodically applied to generate "fresh" references for tracking. However, the frames processed with OD incur large delays, which does not comply with real-time applications requirements. Offloading OD to edge servers can mitigate this issue, but existing work focuses on the optimization of the offloading process in systems where the wireless channel has a very large capacity. Herein, we consider systems with constrained and erratic channel capacity, and establish parallel OT (at the mobile device) and OD (at the edge server) processes that are resilient to large OD latency. We propose Katch-Up, a novel tracking mechanism that improves the system resilience to excessive OD delay. We show that this technique greatly improves the quality of the reference available to tracking, and boosts performance up to 33%. However, while Katch-Up significantly improves performance, it also increases the computing load of the mobile device. Hence, we design SmartDet, a low-complexity controller based on deep reinforcement learning (DRL) that learns to achieve the right trade-off between resource utilization and OD performance. SmartDet takes as input highly-heterogeneous context-related information related to the current video content and the current network conditions to optimize frequency and type of OD offloading, as well as Katch-Up utilization. We extensively evaluate SmartDet on a real-world testbed composed by a JetSon Nano as mobile device and a GTX 980 Ti as edge server, connected through a Wi-Fi link, to collect several network-related traces, as well as energy measurements. We consider a state-of-the-art video dataset (ILSVRC 2015 - VID) and state-of-the-art OD models (EfficientDet 0, 2 and 4). Experimental results show that SmartDet achieves an optimal balance between tracking performance – mean Average Recall (mAR) and resource usage. With respect to a baseline with full Katch-Up usage and maximum channel usage, we still increase mAR by 4% while using 50% less of the channel and 30% power resources associated with Katch-Up. With respect to a fixed strategy using minimal resources, we increase mAR by 20% while using Katch-Up on 1/3 of the frames.

Although mission-critical applications require the use of deep neural networks (DNNs), their continuous execution at mobile devices results in a significant increase in energy consumption. While edge offloading can decrease energy consumption, erratic patterns in channel quality, network and edge server load can lead to severe disruption of the system’s key operations. An alternative approach, called split computing, generates compressed representations within the model (called "bottlenecks"), to reduce bandwidth usage and energy consumption. Prior work has proposed approaches that introduce additional layers, to the detriment of energy consumption and latency. For this reason, we propose a new framework called BottleFit, which, in addition to targeted DNN architecture modifications, includes a novel training strategy to achieve high accuracy even with strong compression rates. We apply BottleFit on cutting-edge DNN models in image classification, and show that BottleFit achieves 77.1% data compression with up to 0.6% accuracy loss on ImageNet dataset, while state of the art such as SPINN loses up to 6% in accuracy. We experimentally measure the power consumption and latency of an image classification application running on an NVIDIA Jetson Nano board (GPU-based) and a Raspberry PI board (GPU-less). We show that BottleFit decreases power consumption and latency respectively by up to 49% and 89% with respect to (w.r.t.) local computing and by 37% and 55% w.r.t. edge offloading. We also compare BottleFit with state-of-the-art autoencoders-based approaches, and show that (i) BottleFit reduces power consumption and execution time respectively by up to 54% and 44% on the Jetson and 40% and 62% on Raspberry PI; (ii) the size of the head model executed on the mobile device is 83 times smaller. We publish the code repository for reproducibility of the results in this study.

Fifth-generation (5G) systems will extensively employ radio access network (RAN) softwarization. This key innovation enables the instantiation of "virtual cellular networks" running on different slices of the shared physical infrastructure. In this paper, we propose the concept of Private Cellular Connectivity as a Service (PCCaaS), where infrastructure providers deploy covert network slices known only to a subset of users. We then present SteaLTE as the first realization of a PCCaaS-enabling system for cellular networks. At its core, SteaLTE utilizes wireless steganography to disguise data as noise to adversarial receivers. Differently from previous work, however, it takes a full-stack approach to steganography, contributing an LTE-compliant stegano-graphic protocol stack for PCCaaS-based communications, and packet schedulers and operations to embed covert data streams on top of traditional cellular traffic (primary traffic). SteaLTE balances undetectability and performance by mimicking channel impairments so that covert data waveforms are almost indistinguishable from noise. We evaluate the performance of SteaLTE on an indoor LTE-compliant testbed under different traffic profiles, distance and mobility patterns. We further test it on the outdoor PAWR POWDER platform over long-range cellular links. Results show that in most experiments SteaLTE imposes little loss of primary traffic throughput in presence of covert data transmissions (<; 6%), making it suitable for undetectable PCCaaS networking.

Due to the sheer scale of the Internet of Things (IoT) and 5G, the wireless spectrum is becoming severely congested. For this reason, wireless devices will need to continuously adapt to current spectrum conditions by changing their communication parameters in real-time. Therefore, wireless signal classification (WSC) will become a compelling necessity to decode fast-changing signals from dynamic transmitters. Thanks to its capability of classifying complex phenomena without explicit mathematical modeling, deep learning (DL) has been demonstrated to be a key enabler of WSC. Although DL can achieve a very high accuracy under certain conditions, recent research has unveiled that the wireless channel can disrupt the features learned by the DL model during training, thus drastically reducing the classification performance in real-world live settings. Since retraining classifiers is cumbersome after deployment, existing work has leveraged the usage of carefully-tailored Finite Impulse Response (FIR) filters that, when applied at the transmitter's side, can restore the features that are lost because of the the channel actions, i.e., waveform synthesis. However, these approaches compute FIRs using offline optimization strategies, which limits their efficacy in highly-dynamic channel settings. In this paper, we improve the state of the art by proposing Chares, a Deep Reinforcement Learning (DRL)-based framework for channel-resilient adaptive waveform synthesis. Chares adapts to new and unseen channel conditions by optimally computing through DRL the FIRs in real time. Chares is a DRL agent whose architecture is based upon the Twin Delayed Deep Deterministic Policy Gradients (TD3), which requires minimal feedback from the receiver and explores a continuous action space for best performance. Chares has been extensively evaluated on two well-known datasets with an extensive number of channels. We have also evaluated the real-time latency of Chares with an implementation on field-programmable gate array (FPGA). Results show that Chares increases the accuracy up to 4.1x when no waveform synthesis is performed, by 1.9x with respect to existing work, and can compute new actions within 41 μs.

Spectrum sharing will be a key technology to tackle spectrum scarcity in the sub-6 GHz bands. To fairly access the shared bandwidth, wireless users will necessarily need to quickly sense large portions of spectrum and opportunistically access unutilized bands. The key unaddressed challenges of spectrum sensing are that (i) it has to be performed with extremely low latency over large bandwidths to detect tiny spectrum holes and to guarantee strict real-time digital signal processing (DSP) constraints; (ii) its underlying algorithms need to be extremely accurate, and flexible enough to work with different wireless bands and protocols to find application in real-world settings. To the best of our knowledge, the literature lacks spectrum sensing techniques able to accomplish both requirements. In this paper, we propose DeepSense, a software/hardware framework for real-time wideband spectrum sensing that relies on real-time deep learning tightly integrated into the transceiver's baseband processing logic to detect and exploit unutilized spectrum bands. DeepSense uses a convolutional neural network (CNN) implemented in the wireless platform's hardware fabric to analyze a small portion of the unprocessed baseband waveform to automatically extract the maximum amount of information with the least amount of I/Q samples. We extensively validate the accuracy, latency and generality performance of DeepSense with (i) a 400 GB dataset containing hundreds of thousands of WiFi transmissions collected "in the wild" with different Signal-to-Noise-Ratio (SNR) conditions and over different days; (ii) a dataset of transmissions collected using our own software-defined radio testbed; and (iii) a synthetic dataset of LTE transmissions under controlled SNR conditions. We also measure the real-time latency of the CNNs trained on the three datasets with an FPGA implementation, and compare our approach with a fixed energy threshold mechanism. Results show that our learning-based approach can deliver a precision and recall of 98% and 97% respectively and a latency as low as 0.61ms. For reproducibility and benchmarking purposes, we pledge to share the code and the datasets used in this paper to the community.

The Long Range (LoRa) protocol for low-power wide-area networks (LPWANs) is a strong candidate to enable the massive roll-out of the Internet of Things (IoT) because of its low cost, impressive sensitivity (-137dBm), and massive scalability potential. As tens of thousands of tiny LoRa devices are deployed over large geographic areas, a key component to the success of LoRa will be the development of reliable and robust authentication mechanisms. To this end, Radio Frequency Fingerprinting (RFFP) through deep learning (DL) has been heralded as an effective zero-power supplement or alternative to energy-hungry cryptography. Existing work on LoRa RFFP has mostly focused on small-scale testbeds and low-dimensional learning techniques; however, many challenges remain. Key among them are authentication techniques robust to a wide variety of channel variations over time and supporting a vast population of devices.

Millions of RFID tags are pervasively used all around the globe to inexpensively identify a wide variety of everyday-use objects. One of the key issues of RFID is that tags cannot use energy-hungry cryptography, and thus can be easily cloned. For this reason, radio fingerprinting (RFP) is a compelling approach that leverages the unique imperfections in the tag's wireless circuitry to achieve large-scale RFID clone detection. Recent work, however, has unveiled that time-varying channel conditions can significantly decrease the accuracy of the RFP process. Prior art in RFID identification does not consider this critical aspect, and instead focuses on custom-tailored feature extraction techniques and data collection with static channel conditions. For this reason, we propose the first large-scale investigation into RFP of RFID tags with dynamic channel conditions. Specifically, we perform a massive data collection campaign on a testbed composed by 200 off-the-shelf identical RFID tags and a software-defined radio (SDR) tag reader. We collect data with different tag-reader distances in an over-the-air configuration. To emulate implanted RFID tags, we also collect data with two different kinds of porcine meat inserted between the tag and the reader. We use this rich dataset to train and test several convolutional neural network (CNN)-based classifiers in a variety of channel conditions. Our investigation reveals that training and testing on different channel conditions drastically degrades the classifier's accuracy. For this reason, we propose a novel training framework based on federated machine learning (FML) and data augmentation (DAG) to boost the accuracy. Extensive experimental results indicate that (i) our FML approach improves accuracy by up to 48%; (ii) our DAG approach improves the FML performance by up to 19% and the single-dataset performance by 31%. To the best of our knowledge, this is the first paper experimentally demonstrating the efficacy of FML and DAG on a large device population. To allow full replicability, we are sharing with the research community our fully-labeled 200-GB RFID waveform dataset, as well as the entirety of our code and trained models, concurrently with our submission.

Highly directional millimeter wave (mmWave) radios need to perform beam management to establish and maintain reliable links. To achieve this objective, existing solutions mostly rely on explicit coordination between the transmitter (TX) and the receiver (RX), which significantly reduces the airtime available for communication and further complicates the network protocol design. This paper advances the state of the art by presenting DeepBeam, a framework for beam management that does not require pilot sequences from the TX, nor any beam sweeping or synchronization from the RX. This is achieved by inferring (i) the Angle of Arrival (AoA) of the beam and (ii) the actual beam being used by the transmitter through waveform-level deep learning on ongoing transmissions between the TX to other receivers. In this way, the RX can associate Signal-to-Noise-Ratio (SNR) levels to beams without explicit coordination with the TX. This is possible because different beam patterns introduce different "impairments" to the waveform, which can be subsequently learned by a convolutional neural network (CNN). To demonstrate the generality of DeepBeam, we conduct an extensive experimental data collection campaign where we collect more than 4 TB of mmWave waveforms with (i) 4 phased array antennas at 60.48 GHz, (ii) 2 codebooks containing 24 one-dimensional beams and 12 two-dimensional beams; (iii) 3 receiver gains; (iv) 3 different AoAs; (v) multiple TX and RX locations. Moreover, we collect waveform data with two custom-designed mmWave software-defined radios with fully-digital beamforming architectures at 58 GHz. We also implement our learning models in FPGA to evaluate latency performance. Results show that DeepBeam (i) achieves accuracy of up to 96%, 84% and 77% with a 5-beam, 12-beam and 24-beam codebook, respectively; (ii) reduces latency by up to 7x with respect to the 5G NR initial beam sweep in a default configuration and with a 12-beam codebook. The waveform dataset and the full DeepBeam code repository are publicly available.

Edge computing enables Mobile Autonomous Systems (MASs) to execute continuous streams of heavy-duty mission-critical processing tasks, such as real-time obstacle detection and navigation. However, in practical applications, erratic patterns in channel quality, network load, and edge server load can interrupt the task flow's execution, which necessarily leads to severe disruption of the system's key operations. Existing work has mostly tackled the problem with reactive approaches, which cannot guarantee task-level reliability. Conversely, in this paper we focus on learning-based predictive edge computing to achieve self-resilient task offloading. By conducting a preliminary experimental evaluation, we show that there is no dominant feature that can predict the edge-MAS system reliability, which calls for an ensemble and selection of weaker features. To tackle the complexity of the problem, we propose SeReMAS, a data-driven optimization framework. We first mathematically formulate a Redundant Task Offloading Problem (RTOP), where a MAS may connect to multiple edge servers for redundancy, and needs to select which server(s) to transmit its computing tasks in order to maximize the probability of task execution while minimizing channel and edge resource utilization. We then create a predictor based on Deep Reinforcement Learning (DRL), which produces the optimum task assignment based on application-, network- and telemetry-based features. We prototype SeReMAS on a testbed composed by a Tarot650 quadcopter drone, mounting a PixHawk flight controller, a Jetson Nano board, and three 802.11n WiFi interfaces. We extensively evaluate SeReMAS by considering an application where one drone offloads high-resolution images for real-time analysis to three edge servers on the ground. Experimental results show that SeReMAS improves the task execution probability by 17% with respect to existing reactive-based approaches. To allow full reproducibility of results, we share the dataset and code with the research community.

Radio access network (RAN) slicing is a virtualization technology that partitions radio resources into multiple autonomous virtual networks. Since RAN slicing can be tailored to provide diverse performance requirements, it will be pivotal to achieve the high-throughput and low-latency communications that next-generation (5G) systems have long yearned for. To this end, effective RAN slicing algorithms must (i) partition radio resources so as to leverage coordination among multiple base stations and thus boost network throughput; and (ii) reduce interference across different slices to guarantee slice isolation and avoid performance degradation. The ultimate goal of this paper is to design RAN slicing algorithms that address the above two requirements. First, we show that the RAN slicing problem can be formulated as a 0-1 Quadratic Programming problem, and we prove its NP-hardness. Second, we propose an optimal solution for small-scale 5G network deployments, and we present three approximation algorithms to make the optimization problem tractable when the network size increases. We first analyze the performance of our algorithms through simulations, and then demonstrate their performance through experiments on a standard-compliant LTE testbed with 2 base stations and 6 smartphones. Our results show that not only do our algorithms efficiently partition RAN resources, but also improve network throughput by 27% and increase by 2× the signal-to-interference-plus-noise ratio.

Deep learning can be used to classify waveform characteristics (e.g., modulation) with accuracy levels that are hardly attainable with traditional techniques. Recent research has demonstrated that one of the most crucial challenges in wireless deep learning is to counteract the channel action, which may significantly alter the waveform features. The problem is further exacerbated by the fact that deep learning algorithms are hardly re-trainable in real time due to their sheer size. This paper proposesDeepFIR, a framework to counteract the channel action in wireless deep learning algorithmswithout retraining the underlying deep learning model. The key intuition is that through the application of a carefully-optimized digital finite input response filter (FIR) at the transmitter’s side, we can apply tiny modifications to the waveform to strengthen its features according to the current channel conditions. We mathematically formulate theWaveform Optimization Problem(WOP)as the problem of finding the optimum FIR to be used on a waveform to improve the classifier’s accuracy. We also propose a data-driven methodology to train the FIRs directly with dataset inputs. We extensively evaluateDeepFIRon an experimental testbed of 20 software-defined radios, as well as on two datasets made up by 500 ADS-B devices and by 500 WiFi devices and a 24-class modulation dataset. Experimental results show that our approach (i) increases the accuracy of the radio fingerprinting models by about 35%, 50% and 58%; (ii) decreases an adversary’s accuracy by about 54% when trying to imitate other device’s fingerprints by using their filters; (iii) achieves 27% improvement over the state of the art on a 100-device dataset; (iv) increases by$2\times$the accuracy of the modulation dataset.

Radio fingerprinting uniquely identifies wireless devices by leveraging tiny hardware-level imperfections inevitably present in off-the-shelf radio circuitry. This way, devices can be directly identified at the physical layer by analyzing the unprocessed received waveform - thus avoiding energy-expensive upper-layer cryptography that resource-challenged embedded devices may not be able to afford. Recent advances have proven that convolutional neural networks (CNNs) - thanks to their multidimensional mappings - can achieve fingerprinting accuracy levels impossible to achieve by traditional low-dimensional algorithms. The same research, however, has also suggested that the wireless channel may negatively impact the accuracy of CNN-based radio fingerprinting algorithms by making device-unique hardware imperfections much harder to recognize.In spite of the growing interest in radio fingerprinting research by academia and DARPA, the wireless research community still lacks (i) a large-scale open dataset for radio fingerprinting collected in diverse environments and rich, diverse, channel conditions; and (ii) a full-fledged, systematic, quantitative investigation of the impact of the wireless channel on the accuracy of CNN-based radio fingerprinting algorithms. The key contribution of this paper is to bridge this gap by (i) collecting and sharing with the community more than 7TB of wireless data obtained from 20 wireless devices with identical RF circuitry (and thus, worst-case scenario for fingerprinting) over the course of several days in (a) an anechoic chamber, (b) in-the-wild testbed, and (c) with cable connections; and (ii) providing a first-of-its-kind evaluation of the impact of the wireless channel on CNN-based fingerprinting algorithms through (a) the 7TB experimental dataset and (b) a 400GB dataset provided by DARPA containing hundreds of thousands of transmissions from thousands of WiFi and ADS-B devices with different SNR conditions. Experimental results conclude that (i) the wireless channel impacts the classification accuracy significantly, i.e., from 85% to 9% and from 30% to 17% in the experimental and DARPA dataset, respectively; and that (ii) equalizing I/Q data can increase the accuracy to a significant extent (i.e., by up to 23%) when the number of devices increases significantly.

Recent work has demonstrated that cutting-edge advances in deep reinforcement learning (DRL) may be leveraged to empower wireless devices with the much-needed ability to "sense" current spectrum and network conditions and "react" in real time by either exploiting known optimal actions or exploring new actions. Yet, understanding whether real-time DRL can be at all applied in the resource-challenged embedded IoT domain, as well as designing IoT-tailored DRL systems and architectures, still remains mostly uncharted territory. This paper bridges the existing gap between the extensive theoretical research on wireless DRL and its system-level applications by presenting Deep Wireless Embedded Reinforcement Learning (DeepWiERL), a general-purpose, hybrid software/hardware DRL framework specifically tailored for embedded IoT wireless devices. DeepWiERL provides abstractions, circuits, software structures and drivers to support the training and real-time execution of state-of-the-art DRL algorithms on the device's hardware. Moreover, DeepWiERL includes a novel supervised DRL model selection and bootstrap (S-DMSB) technique that leverages transfer learning and high-level synthesis (HLS) circuit design to orchestrate a neural network architecture that satisfies hardware and application throughput constraints and speeds up the DRL algorithm convergence. Experimental evaluation on a fully-custom software-defined radio testbed (i) proves for the first time the feasibility of real-time DRL-based algorithms on a real-world wireless platform with multiple channel conditions; (ii) shows that DeepWiERL supports 16x data rate and consumes 14x less energy than a software-based implementation, and (iii) indicates that S-DMSB may improve the DRL convergence time by 6x and increase the obtained reward by 45% if prior channel knowledge is available.

Network slicing of multi-access edge computing (MEC) resources is expected to be a pivotal technology to the success of 5G networks and beyond. The key challenge that sets MEC slicing apart from traditional resource allocation problems is that edge nodes depend on tightly-intertwined and strictly-constrained networking, computation and storage resources. Therefore, instantiating MEC slices without incurring in resource over-provisioning is hardly addressable with existing slicing algorithms. The main innovation of this paper is Sl-EDGE, a unified MEC slicing framework that allows network operators to instantiate heterogeneous slice services (e.g., video streaming, caching, 5G network access) on edge devices. We first describe the architecture and operations of Sl-EDGE, and then show that the problem of optimally instantiating joint network-MEC slices is NP-hard. Thus, we propose near-optimal algorithms that leverage key similarities among edge nodes and resource virtualization to instantiate heterogeneous slices 7.5x faster and within 25% of the optimum. We first assess the performance of our algorithms through extensive numerical analysis, and show that Sl-EDGE instantiates slices 6x more efficiently then state-of-the-art MEC slicing algorithms. Furthermore, experimental results on a 24-radio testbed with 9 smartphones demonstrate that Sl-EDGE provides simultaneously highly-efficient slicing of joint LTE connectivity, video streaming over WiFi, and ffmpeg video transcoding.

Today's wireless technologies are largely based on inflexible designs, which makes them inefficient and prone to a variety of wireless attacks. To address this key issue, wireless receivers will need to (i) infer on-the-fly the physical-layer parameters currently used by transmitters; and if needed, (ii) change their hardware and software structures to demodulate the incoming waveform. In this paper, we introduce PolymoRF, a deep learning-based polymorphic receiver able to reconfigure itself in real time based on the inferred waveform parameters. Our key technical innovations are (i) a novel embedded deep learning architecture, called RFNet, which enables the solution of key waveform inference problems; (ii) a generalized hardware/software architecture that integrates RFNet with radio components and signal processing. We prototype PolymoRF on a custom software-defined radio platform, and show through extensive over-the-air experiments that PolymoRF achieves throughput within 87% of a perfect-knowledge Oracle system, thus demonstrating for the first time that polymorphic receivers are feasible.

The extremely high data rates provided by communications in the millimeter-length (mmWave) frequency bands can help address the unprecedented demands of next-generation wireless communications. However, atmospheric attenuation and high propagation loss severely limit the coverage of mmWave networks. To overcome these challenges, multi-input-multi-output (MIMO) provides beamforming capabilities and high-gain steerable antennas to expand communication coverage at mmWave frequencies. The main contribution of this paper is the performance evaluation of mmWave communications on top of the recently released NR standard for 5G cellular networks. Furthermore, we compare the performance of NR with the 4G long-term evolution (LTE) standard on a highly realistic campus environment. We consider physical layer constraints such as transmit power, ambient noise, receiver noise figure, and practical antenna gain in both cases, and examine bitrate and area coverage as the criteria to benchmark the performance. We also show the impact of MIMO technology to improve the performance of the 5G NR cellular network. Our evaluation demonstrates that 5G NR provides on average 6.7 times bitrate improvement without remarkable coverage degradation.

Undetectable wireless transmissions are fundamental to avoid eavesdroppers or censorship by authoritarian governments. To address this issue, wireless steganography “hides” covert information inside primary information by slightly modifying the transmitted waveform such that primary information will still be decodable, while covert information will be seen as noise by agnostic receivers. Since the addition of covert information inevitably decreases the SNR of the primary transmission, a key challenge in wireless steganography is to mathematically analyze and optimize the impact of the covert channel on the primary channel as a function of different channel conditions. Another core issue is to make sure that the covert channel is almost undetectable by eavesdroppers. Existing approaches are protocol-specific and thus their performance cannot be assessed and optimized in general scenarios. To address this research gap, we notice that existing wireless technologies rely on phase-keying modulations (e.g., BPSK, QPSK) that in most cases do not use the channel up to its Shannon capacity. Therefore, the residual capacity can be leveraged to implement a wireless system based on a pseudo-noise asymmetric shift keying (PN-ASK) modulation, where covert symbols are mapped by shifting the amplitude of primary symbols. This way, covert information will be undetectable, since a receiver expecting phase-modulated symbols will see their shift in amplitude as an effect of channel/path loss degradation. Through rigorous mathematical analysis, we first investigate the SER of PN-ASK as a function of the channel; then, we find the optimal PN-ASK parameters that optimize primary and covert throughput under different channel condition. We evaluate the throughput performance and undetectability of PN-ASK through extensive simulations and on an experimental testbed based on USRP N210 software-defined radios. Results indicate that PN-ASK improves the throughput by more than 8x with respect to prior art. Finally, we demonstrate through experiments that PN-ASK is able to transmit covert data on top of IEEE 802.11g frames, which are correctly decoded by an off-the-shelf laptop WiFi card without any hardware modifications.

The notions of softwarization and virtualization of the radio access network (RAN) of next-generation (5G) wireless systems are ushering in a vision where applications and services are physically decoupled from devices and network infrastructure. This crucial aspect will ultimately enable the dynamic deployment of heterogeneous services by different network operators over the same physical infrastructure. RAN slicing is a form of 5G virtualization that allows network infrastructure owners to dynamically “slice” and “serve” their network resources (i. e., spectrum, power, antennas, among others) to different mobile virtual network operators (MVNOs), according to their current needs. Once the slicing policy (i.e., the percentage of resources assigned to each MVNO) has been computed, a major challenge is how to allocate spectrum resources to MVNOs in such a way that (i) the slicing policy defined by the network owner is enforced; and (ii) the interference among different MVNOs is minimized. In this article, we mathematically formalize the RAN slicing enforcement problem (RSEP) and demonstrate its NP-hardness. For this reason, we design three approximation algorithms that render the solution scalable as the RSEP increases in size. We extensively evaluate their performance through simulations and experiments on a testbed made up of 8 software-defined radio peripherals. Experimental results reveal that not only do our algorithms enforce the slicing policies, but can also double the total network throughput when intra-MVNO power control policies are used in conjunction.

The explosion of 5G networks and the Internet of Things will result in an exceptionally crowded RF environment, where techniques such as spectrum sharing and dynamic spectrum access will become essential components of the wireless communication process. In this vision, wireless devices must be able to (i) learn to autonomously extract knowledge from the spectrum on-the-fly; and (ii) react in real time to the inferred spectrum knowledge by appropriately changing communication parameters, including frequency band, symbol modulation, coding rate, among others. Traditional CPU-based machine learning suffers from high latency, and requires application-specific and computationally-intensive feature extraction/selection algorithms. Conversely, deep learning allows the analysis of massive amount of unprocessed spectrum data without ad-hoc feature extraction. So far, deep learning has been used for offline wireless spectrum analysis only. Therefore, additional research is needed to design systems that bring deep learning algorithms directly on the device's hardware and tightly intertwined with the RF components to enable real-time spectrum-driven decision-making at the physical layer. In this paper, we present RFLearn, the first system enabling spectrum knowledge extraction from unprocessed I/Q samples by deep learning directly in the RF loop. RFLearn provides (i) a complete hardware/software architecture where the CPU, radio transceiver and learning/actuation circuits are tightly connected for maximum performance; and (ii) a learning circuit design framework where the latency vs. hardware resource consumption trade-off can explored. We implement and evaluate the performance of RFLearn on custom software-defined radio built on a system-on-chip (SoC) ZYNQ-7000 device mounting AD9361 radio transceivers and VERT2450 antennas. We showcase the capabilities of RFLearn by applying it to solving the fundamental problems of modulation and OFDM parameter recognition. Experimental results reveal that RFLearn decreases latency and power by about 17x and 15x with respect to a software-based solution, with a comparatively low hardware resource consumption.

Radio fingerprinting provides a reliable and energy-efficient IoT authentication strategy by leveraging the unique hardware-level imperfections imposed on the received wireless signal by the transmitter's radio circuitry. Most of existing approaches utilize hand-tailored protocol-specific feature extraction techniques, which can identify devices operating under a pre-defined wireless protocol only. Conversely, by mapping inputs onto a very large feature space, deep learning algorithms can be trained to fingerprint large populations of devices operating under any wireless standard.

Recent research advances in wireless security have shown that advanced jamming can significantly decrease the performance of wireless communications. In advanced jamming, the adversary intentionally concentrates the available energy budget on specific critical components (e.g., pilot symbols, acknowledgement packets, etc.) to (i) increase the jamming effectiveness, as more targets can be jammed with the same energy budget; and (ii) decrease the likelihood of being detected, as the channel is jammed for a shorter period of time. These key aspects make advanced jamming very stealthy yet exceptionally effective in practical scenarios. One of the fundamental challenges in designing defense mechanisms against an advanced jammer is understanding which jamming strategies yields the lowest throughput, for a given channel condition and a given amount of energy. To the best of our knowledge, this problem still remains unsolved, as an analytic model to quantitatively compare advanced jamming schemes is still missing in existing literature. To fill this gap, in this paper we conduct a comparative analysis of several most viable advanced jamming schemes in the widely-used MIMO networks. We first mathematically model a number of advanced jamming schemes at the signal processing level, so that a quantitative relationship between the jamming energy and the jamming effect is established. Based on the model, theorems are derived on the optimal advanced jamming scheme for an arbitrary channel condition. The theoretical findings are validated through extensive simulations and experiments on a 5-radio 2x2 MIMO testbed. Our results show that the theorems are able to predict jamming efficiency with high accuracy. Moreover, to further demonstrate that the theoretical findings are applicable to address crucial real-world jamming problems, we show that the theorems can be incorporated to state-of-art reinforcement-learning based jamming algorithms and boost the action exploration phase so that a faster convergence is achieved.

The promise of real-time detection and response to life-crippling diseases brought by the Implantable Internet of Medical Things (IIoMT) has recently spurred substantial advances in implantable technologies. Yet, existing devices do not provide at once the miniaturized end-to-end sensing-computation-communication-recharging capabilities to implement IIoMT applications. This paper fills the existing research gap by presenting U-Verse, the first FDA-compliant rechargeable IIoMT platform packing sensing, computation, communication, and recharging circuits into a penny-scale platform. U-Verse uses a single miniaturized transducer for data exchange and for wireless charging. To predict U-Verse's performance, we (i) derive and experimentally validate a mathematical model of U-Verse's charging efficiency; and (ii) experimentally calculate the resistance-reactance parameters of our ultrasonic transducer and rectifying circuit. We design a matching circuit to maximize the amount of power transferred from the outside. We also go through the challenge of fabricating a full-fledged cm-scale printed circuit board (PCB) for U-Verse. Extensive experimental evaluation indicates that U-Verse (i) is able to recharge a 330mF and 15F energy storage unit - several orders of magnitude higher than existing work - respectively under 20 and 60 minutes at a depth of 5cm; (ii) achieves stored charge duration of up to 610 and 40 hours in case of battery and supercapacitor energy storage, respectively. Finally, U-Verse is demonstrated through (i) a closed-loop application where a periodic sensing/actuation task sends data via ultrasounds through real porcine meat; and (ii) a real-time reconfigurable pacemaker.

Mobile crowdsensing harnesses the sensing power of modern smartphones to collect and analyze data beyond the scale of what was previously possible with traditional sensor networks. Given the participatory nature of mobile crowdsensing, it is imperative to incentivize mobile users to provide sensing services in a timely and reliable manner. Most importantly, given sensed information is often valid for a limited period of time, the capability of smartphone users to execute sensing tasks largely depends on their mobility pattern, which is often uncertain. For this reason, in this paper, we propose IncentMe, a framework that solves this core issue by leveraging game-theoretical reverse auction mechanism design. After demonstrating that the proposed problem is NP-hard, we derive two mechanisms that are parallelizable and achieve higher approximation ratio than existing work. IncentMe has been extensively evaluated on a road traffic monitoring application implemented using mobility traces of taxi cabs in San Francisco, Rome, and Beijing. Results demonstrate that the mechanisms in IncentMe outperform the state of the art work by improving the efficiency in recruiting participants by 30 percent.

Wireless networks are extremely vulnerable to a plethora of security threats, including eavesdropping, jamming, and spoofing, to name a few. Recently, a number of next-generation cross-layer attacks have been unveiled, which leverage small changes on one network layer to stealthily and significantly compromise another target layer. Since cross-layer attacks are stealthy, dynamic, and unpredictable in nature, novel security techniques are needed. Since models of the environment and attacker's behavior may be hard to obtain in practical scenarios, machine learning techniques become the ideal choice to tackle cross-layer attacks. In this paper, we propose FORMAT, a novel framework to tackle cross-layer security attacks in wireless networks. FORMAT is based on Bayesian learning and made up by a detection and a mitigation component. On one hand, the attack detection component constructs a model of observed evidence to identify stealthy attack activities. On the other hand, the mitigation component uses optimization theory to achieve the desired trade-off between security and performance. The proposed FORMAT framework has been extensively evaluated and compared with existing work by simulations and experiments obtained with a real-world testbed made up by Ettus Universal Software Radio Peripheral (USRP) radios. Results demonstrate the effectiveness of the proposed methodology as FORMAT is able to effectively detect and mitigate the considered cross-layer attacks.

Thanks to the collective action of participating smartphone users, mobile crowdsensing allows data collection at a scale and pace that was once impossible. The biggest challenge to overcome in mobile crowdsensing is that participants may exhibit malicious or unreliable behavior, thus compromising the accuracy of the data collection process. Therefore, it becomes imperative to design algorithms to accurately classify between reliable and unreliable sensing reports. To address this crucial issue, we propose a novel Framework for optimizing Information Reliability in Smartphone-based participaTory sensing (FIRST) that leverages mobile trusted participants (MTPs) to securely assess the reliability of sensing reports. FIRST models and solves the challenging problem of determining before deployment the minimum number of MTPs to be used to achieve desired classification accuracy. After a rigorous mathematical study of its performance, we extensively evaluate FIRST through an implementation in iOS and Android of a room occupancy monitoring system and through simulations with real-world mobility traces. Experimental results demonstrate that FIRST reduces significantly the impact of three security attacks (i.e., corruption, on/off, and collusion) by achieving a classification accuracy of almost 80% in the considered scenarios. Finally, we discuss our ongoing research efforts to test the performance of FIRST as part of the National Map Corps project.

The Internet of Things (IoT) realizes a vision where billions of interconnected devices are deployed just about everywhere, from inside our bodies to the most remote areas of the globe. As the IoT will soon pervade every aspect of our lives and will be accessible from anywhere, addressing critical IoT security threats is now more important than ever. Traditional approaches where security is applied as an afterthought and as a “patch” against known attacks are insufficient. Indeed, next-generation IoT challenges will require a new secure-by-design vision, where threats are addressed proactively and IoT devices learn to dynamically adapt to different threats. To this end, machine learning (ML) and software-defined networking (SDN) will be key to provide both reconfigurability and intelligence to the IoT devices. In this paper, we first provide a taxonomy and survey the state of the art in IoT security research, and offer a roadmap of concrete research challenges related to the application of ML and SDN to address existing and next-generation IoT security threats.

Radio access network (RAN) slicing is an effective methodology to dynamically allocate networking resources in 5G networks. One of the main challenges of RAN slicing is that it is provably an NP-Hard problem. For this reason, we design near-optimal low-complexity distributed RAN slicing algorithms. First, we model the slicing problem as a congestion game, and demonstrate that such game admits a uniqueNash equilibrium(NE). Then, we evaluate thePrice of Anarchy(PoA) of the NE, i.e., the efficiency of the NE as compared with the social optimum, and demonstrate that the PoA is upper-bounded by 3/2. Next, we propose two fully-distributed algorithms that provably converge to the unique NE without revealing privacy-sensitive parameters from the slice tenants. Moreover, we introduce an adaptive pricing mechanism of the wireless resources to improve the network owner’s profit. We evaluate the performance of our algorithms through simulations and an experimental testbed deployed on the Amazon EC2 cloud, both based on a real-world dataset of base stations from the OpenCellID project. Results conclude that our algorithms converge to the NE rapidly and achieve near-optimal performance, while our pricing mechanism effectively improves the profit of the network owner.

Smartphones have become the most pervasive devices in people’s lives and are clearly transforming the way we live and perceive technology. Today’s smartphones benefit from almost ubiquitous Internet connectivity and come equipped with a plethora of inexpensive yet powerful embedded sensors, such as an accelerometer, a gyroscope, a microphone, and a camera. This unique combination has enabled revolutionary applications based on the mobile crowdsensing paradigm, such as real-time road traffic monitoring, air and noise pollution, crime control, and wildlife monitoring, just to name a few. Differently from prior sensing paradigms, humans are now the primary actors of the sensing process, since they become fundamental in retrieving reliable and up-to-date information about the event being monitored. As humans may behave unreliably or maliciously, assessing and guaranteeing Quality of Information (QoI) becomes more important than ever. In this article, we provide a new framework for defining and enforcing the QoI in mobile crowdsensing and analyze in depth the current state of the art on the topic. We also outline novel research challenges, along with possible directions of future work.

The reliability of location-based services (LBS) is strongly dependent on the accuracy of the location of the users. However, existing LBS systems are not able to efficiently validate the position of users in large-scale outdoor environments, leading to possible location spoofing attacks by malicious users. To this end, we present an efficient and scalable Location Validation System (LVS) that secures LBS systems from location spoofing attacks. In particular, the user location is verified with the help of mobile WiFi hotspots (MHSs), who are users activating the WiFi hotspot capability of their smartphones and accept connections from nearby users, thereby validating their position inside the sensing area. The system also comprises a novel verification technique called Chains of Sight, which tackles collusion-based attacks effectively. LVS also includes a reputation-based algorithm that rules out sensing reports of location-spoofing users.

In emergency scenarios such as earthquakes, fires, avalanches, or building collapses, it is necessary to discover people trapped under debris or anyway hidden from eyesight. In this paper, we propose RescuePal, an energy-efficient smartphone-based system that does not require any interaction by the victim and does not use energy-expensive GPS. RescuePal leverages a wake-up system based on sounds that activates the WiFi interface of the victim's smartphone only when the rescuer is close, to save energy. After presenting the system, we mathematically formulate an optimization problem so as to find the sound frequency and power level that minimizes WiFi false activations and yet guarantees high discovery efficiency. RescuePal has been implemented on off-the-shelf Android-based devices, and its performance has been evaluated on a realistic use-case scenario of victims inside a building. Finally, the energy consumption of RescuePal has been calculated using the Power Monitor hardware tool. Results demonstrate that RescuePal is highly effective and saves more than 60% of energy with respect to an approach based only on WiFi.

Many analytical models have been proposed for evaluating the performance of event-driven 802.15.4 Wireless Sensor Networks (WSNs), in Non-Beacon Enabled (NBE) mode. However, existing models do not provide accurate analysis of large-scale WSNs, due to tractability issues and/or simplifying assumptions. In this paper, we propose a new approach called Event Chains Computation (ECC) to model the unslotted CSMA/CA algorithm used for channel access in NBE mode. ECC relies on the idea that outcomes of the CSMA/CA algorithm can be represented as chains of events that subsequently occur in the network. Although ECC can generate all the possible outcomes, it only considers chains with a probability to occur greater than a pre-defined threshold to reduce complexity. Furthermore, ECC parallelizes the computation by managing different chains through different threads. Our results show that, by an appropriate threshold selection, the time to derive performance metrics can be drastically reduced, with negligible impact on accuracy. We also show that the computation time decreases almost linearly with the number of employed threads. We validate our model through simulations and testbed experiments, and use it to investigate the impact of different parameters on the WSN performance, in terms of delivery ratio, latency, and energy consumption.

In past literature, it has been demonstrated that the use of mobile sinks (MSs) increases dramatically the lifetime of wireless sensor networks (WSNs). In applications where the MSs are humans, animals, or transportation systems, the mobility of the MSs is often uncontrollable and could also be random and unpredictable. This implies the necessity of algorithms tailored to handle uncertainty on the MS mobility. In this article, we define the lifetime optimization of a WSN in the presence of uncontrollable sink mobility and Quality of Service (QoS) constraints. After defining an ideal scheme (calledOracle) which provably maximizes network lifetime, we present a novelSwarm-Intelligence-based Sensor Selection Algorithm(SISSA), which optimizes network lifetime and meets predefined QoS constraints. Then we mathematically analyze SISSA and derive analytical bounds on energy consumption, number of messages exchanged, and convergence time. The algorithm is experimentally evaluated on practical experimental setups, and its performances are compared to that by the optimalOraclescheme, as well as with the IEEE 802.15.4 MAC and TDMA schemes. Results conclude that SISSA provides on the average the 56% of the lifetime provided byOracleand outperforms IEEE 802.15.4 and TDMA in terms of yielded network lifetime.

Participatory sensing is a powerful paradigm that takes advantage of smartphones to collect and analyze data beyond the scale of what was previously possible. Given that participatory sensing systems rely completely on the users’ willingness to submit up-to-date and accurate information, it is paramount to effectively incentivize users’ active and reliable participation. In this article, we survey existing literature on incentive mechanisms for participatory sensing systems. In particular, we present a taxonomy of existing incentive mechanisms for participatory sensing systems, which are subsequently discussed in depth by comparing and contrasting different approaches. Finally, we discuss an agenda of open research challenges in incentivizing users in participatory sensing.

In past literature, it has been demonstrated that the use of mobile sinks (MSs) increases dramatically the lifetime of wireless sensor networks (WSNs). In applications where the MSs are humans, animals, or transportation systems, the mobility of the MS is often random and unpredictable, implying the necessity of novel and specific algorithms able to deal with large uncertainty on the MS mobility. In this paper, we define the yet unsolved problem of optimizing the lifetime of a WSN in the presence of uncontrollable and random sink mobility with QoS constraints. Then, we present a novel Swarm-Intelligence-based Sensor Selection Algorithm (SISSA), which optimizes network lifetime and meets pre-defined QoS constraints. Next, we mathematically analyze SISSA and derive analytical bounds on energy consumption, number of messages exchanged, and convergence time. The efficiency of SISSA and the accuracy of the model are experimentally evaluated with a testbed composed by 40 sensors, and the network lifetime provided by SISSA is compared to that by an ideal scheme. Experimental and analytical results conclude that SISSA is highly scalable and energy-efficient, and provides on the average the 56% of the lifetime provided by the ideal scheme in all the considered network parameter sets.

Participatory sensing (PS) has recently attracted tremendous attention given its potential for a wide variety of sensing applications. Due to the fact that PS systems rely completely on the data provided by the users, incentivizing users' active participation while guaranteeing data reliability is paramount to effectively employ PS systems in practical scenarios. In this paper, we first define a set of attacks which compromise data reliability of existing PS applications. Next, we propose a scalable and secure trust-based framework, called FIDES, which relies on the concept of mobile security agents (MSAs) and Josang's trust model to rule out incorrect reports and reward reliable users. By simulating the FIDES framework on mobility traces of taxi cabs in San Francisco, we demonstrate that FIDES secures the PS system from the proposed attacks, guarantees high data reliability, and saves significant amount of revenue with respect to existing reward mechanisms.

Recent studies have demonstrated that mobile elements (MEs) are an efficient solution to help decrease dramatically energy consumption in wireless sensor networks (WSNs). However, in most of cases, sensors use duty cycle schemes to save energy, and unless the ME mobility pattern is deterministic, each sensor node has to discover the presence of the ME in the nearby area before starting to exchange data with it. Therefore, in such wireless sensor networks with mobile elements (in short, WSN-MEs), the definition and analysis of a protocol for efficient ME discovery becomes of fundamental importance. In this paper, we propose an extensive performance analysis of an easy-to-implement, hierarchical discovery protocol for WSN-MEs, called Dual Beacon Discovery (2BD) protocol, taking into account stochastic, multi-path, variable speed ME mobility patterns. We also derive the optimal parameter values that minimize the energy consumption of sensor nodes, while guaranteeing the minimum node throughput required by the applications under consideration. Finally, we compare the 2BD protocol with a classical solution based on Periodic Listening (PL). Our results show that 2BD can exploit its hierarchical mechanism and thus significantly increase lifetime, especially when the ME discovery phase is relatively long.

In sparse wireless sensor networks, data collection is carried out through specialized mobile nodes that visit sensor nodes, gather data, and transport them to the sink node. Since visit times are typically unpredictable, one of the main challenges to be faced in this kind of networks is the energy-efficient discovery of mobile collector nodes by sensor nodes. In this paper, we propose an adaptive discovery algorithm that combines a learning-based approach with a hierarchical scheme. Thanks to its hybrid nature, the proposed algorithm is very flexible, as it can adapt to very different mobility patterns of the mobile collector node(s), ranging from deterministic to completely random mobility. We have investigated the performance of the proposed approach, through simulation, and we have compared it with existing adaptive algorithms that only leverage either a learning-based or a hierarchical approach. Our results show that the proposed hybrid algorithm outperforms the considered adaptive approaches in all the analyzed scenarios.

Wireless Sensor Networks (WSNs) are emerging as an effective solution for a wide range of real-life applications. In scenarios where a fine-grain sensing is not required, sensor nodes can be sparsely deployed in strategic locations and special Mobile Elements (MEs) can be used for data collection. Since communication between a sensor node and a ME can occur only when they are in the transmission range of each other, one of the main challenges in the design of a WSN with MEs is the energy-efficient and timely discovery of MEs. In this paper, we consider a hierarchical ME discovery protocol, namely Dual beacon Discovery (2BD) protocol, based on two different beacon messages emitted by the ME (i.e., Long-Range Beacons and Short-Range Beacons). We develop a detailed analytical model of 2BD assuming a sparse network scenario, and derive the optimal parameter values that minimize the energy consumption at sensor nodes, while guaranteeing the minimum throughput required by the application. Finally, we compare the energy efficiency and performance of 2BD with those of a traditional discovery protocol based on a single beacon. Our results show that 2BD can provide significant energy savings, especially when the discovery phase is relatively long.