Cryptographic wallets play a vital role in securing digital assets within blockchain networks by managing private keys that authorize secure transactions. However, side channel analysis (SCA) attacks have become a serious threat, enabling attackers to extract sensitive information by exploiting algorithmic weaknesses in microcontroller-based wallets, resulting in the loss of millions of dollars in digital assets. In hierarchically deterministic (HD) systems, the compromise of a single primary key can endanger all subsequent child keys, while the use of independent keys for each account introduces complexity and challenges in key management. This work presents HardVault, a field programmable gate array (FPGA)-based cryptocurrency wallet that supports both Bitcoin and Ethereum. HardVault introduces the first hardware wallet architecture that implements both non-deterministic (ND) and HD key generation modes directly in hardware, giving users the flexibility to choose either approach based on their security and usability needs. By leveraging constant-time operations and hardware-enforced private-key isolation, the design significantly improves resilience to SCA attacks. In addition, the architecture prioritizes resource efficiency to minimize area usage without compromising security, making it well-suited for compact, portable hardware wallet applications. Implementation on a ZCU104 FPGA shows that HardVault uses only 27% of available look-up tables (LUTs). Compared to the Trezor One cryptocurrency (crypto) wallet, the proposed implementation achieves$9\times $higher energy efficiency,$8\times $lower latency, and$7\times $higher throughput.

Ethereum leverages smart contracts (SCs) to power decentralized applications (dApps), with execution handled by the Ethereum virtual machine (EVM) within an Ethereum client. Other blockchain platforms, including Avalanche, Polkadot, Aurora, and Cardano, have also adopted the EVM. However, the performance of the EVM is often constrained by the limitations of general-purpose processors, a challenge that has been explored in the literature. This work aims to further address the limitation by proposing EVMx, a dedicated single-core SC execution engine implemented on a field programmable gate array (FPGA). EVMx follows a processor-like architecture inspired by the RISC philosophy. By exploiting the parallelism and high-speed processing capabilities of FPGA hardware, EVMx achieves a 61% to 99% reduction in execution time for commonly used operation codes compared to traditional central processing unit (CPU)-based environments. Furthermore, EVMx executes entire Ethereum blocks with a percentage reduction in execution time between 6% and 56% against comparable FPGA implementations and 98% to 99% compared to CPU-based EVMs in the literature. These results demonstrate the potential of EVMx to significantly accelerate SC execution and enhance the performance of EVM-compatible blockchains.

Ethereum blockchain uses smart contracts (SCs) to implement decentralized applications (dApps). SCs are executed by the Ethereum virtual machine (EVM) running within an Ethereum client. Moreover, the EVM has been widely adopted by other blockchain platforms, including Solana, Cardano, Avalanche, Polkadot, and more. However, the EVM performance is limited by the constraints of the general-purpose computer it operates on. This work proposes offloading SC execution onto a dedicated hardware-based EVM. Specifically, EVMx is an FPGA-based SC execution engine that benefits from the inherent parallelism and high-speed processing capabilities of a hardware architecture. Synthesis results demonstrate a reduction in execution time of 72% to 99% for commonly used operation codes compared to CPU-based SC execution environments. Moreover, the execution time of Ethereum blocks on EVMx is up to 6 ×faster compared to analogous works in the literature. These results highlight the potential of the proposed architecture to accelerate SC execution and enhance the performance of EVM-compatible blockchains.

The widespread adoption of IoT networks has made them vulnerable to jamming attacks, which disrupt communication and compromise critical applications. Traditional jamming detection methods face challenges due to resource constraints and the need for large labeled datasets. This paper proposes a Deep Active Learning (DAL) framework for jamming detection and classification, addressing these limitations by minimizing the reliance on annotated data while maintaining high accuracy. Leveraging pool-based and uncertainty sampling, our approach iteratively selects the most informative data points for labeling, significantly reducing the dataset size required for training. Using a real-world IoT dataset, the proposed framework achieved 98% accuracy with only 50% of the available dataset. This represents a 13% improvement in accuracy and a 50% reduction in dataset size. Experimental results, validated through Monte Carlo simulations, demonstrated the model’s robustness in distinguishing between normal channels, constant jammers, and periodic jammers, with minimal misclassification. The framework’s efficiency and accuracy make it a practical solution for resource-constrained IoT networks.

The development of beyond-fifth-generation (B5G) communication systems introduces challenges in maintaining timing and frequency synchronization, especially in low SNR and extended coverage scenarios. Accurate carrier frequency offset (CFO) estimation is crucial for establishing calls under such conditions. Existing methods, like maximum likelihood estimation, have limitations, while machine learning (ML) techniques have shown promise in wireless communication. In this work, we propose an ML-based approach using Long Short-Term Memory (LSTM) neural networks and automated machine learning (AutoML) to tune hyperparameters and improve CFO estimation accuracy. We compare our model with a gradientboosting machine (GBM) approach and demonstrate superior accuracy. Our research addresses CFO estimation challenges in B5G systems and offers valuable insights for the development of robust techniques in advanced communication systems.

Learning to disentangle and represent factors of variation in data is an important problem in artificial intelligence. While many advances have been made to learn these representations, it is still unclear how to quantify disentanglement. While several metrics exist, little is known on their implicit assumptions, what they truly measure, and their limits. In consequence, it is difficult to interpret results when comparing different representations. In this work, we survey supervised disentanglement metrics and thoroughly analyze them. We propose a new taxonomy in which all metrics fall into one of the three families: intervention-based, predictor-based, and information-based. We conduct extensive experiments in which we isolate properties of disentangled representations, allowing stratified comparison along several axes. From our experiment results and analysis, we provide insights on relations between disentangled representation properties. Finally, we share guidelines on how to measure disentanglement.

This work presents a system for measuring and analyzing motion, by a portable electronic device and a flexible fiber sensor. The fiber is composed of multi-walled carbon nanotubes (MWCNTs) for its conductivity and polydimethylsiloxane elastomer (PDMS) for its elasticity. A new sensor interface circuit was designed in this study to interface with the fiber and measure its impedance. The measured impedance data are sampled and transmitted via Bluetooth to a laptop. The characteristics of the fiber and a wireless measurement system allow an easy integration into a smart garment to monitor various vital signs and motion markers (e.g angle, step). The system was assessed on a robotic arm before being put in realistic situations through various exercises (flexion/extension knee movements, standing multi-joint movements and walk/run on treadmill) on 5 participants for its ability to measure angle, number of movements, rate and speed. In addition, fibers measurement endurance capacities over months were observed. An assessment of the fiber impedance measurement circuit was performed (minimum resolution of$25~\Omega $, relative error of 2.82% on the estimated value of resistance). Tests carried out over a period of several months show that the fiber maintained good measurement performance when tested on a robotic arm, given an average correlation of 0.85 between angle and fiber impedance. The relative error (RE) made on the number of detected movements was 6.57% in average. In realistic workout situations, these values respectively reached between 0.58 and 73.95% for flexion/extension knee movements. A correlation factor of 0.76 was obtained when the participants were walking on a treadmill at a given speed. Otherwise, RE on number of movements was 8.33% for treadmill exercise, 12.84% on standing exercise. For these exercises and for the movement rate, the average correlation calculated with the reference was between 0.75 and 0.33. Finally, RE on estimated speed was 23.3% in average for the treadmill exercise. The system (sensor interface circuit and Fiber) allows to properly monitor human motion in various activities.

Massive Machine Type Communications (mMTC) is one of the main features supported by the 3GPP New Radio (NR) technology. The new system specifications render multiple challenges for radio designers. Machine terminals are required to setup a call at signal-to-noise ratio (SNR) as low as -10dB in the extended coverage mode. Further, only one receive antenna is to be used with almost no frequency diversity. Maintaining timing and frequency synchronization with such requirements is a challenging task. In this work, we present, and evaluate the potential timing and frequency offset tracking algorithms subject to the new system requirements. The performance of the presented algorithms can be improved at low SNR regimes by employing the time averaging approach on the account of increasing the processing time. A comparison of the performance of different algorithms using system simulations is presented for different channel conditions.

Millimeter wave (mmWave) massive MIMO systems will be used extensively in future communications systems to provide high data rates. For such systems, hybrid precoders are preferred to fully digital precoders in decreasing the cost and energy consumption. In this paper, we propose a partially connected hybrid precoding design on the orthogonal matching pursuit (OMP) algorithm. The proposed algorithm accounts for the limitations of analog beamforming circuitry and assumes channel state information (CSI) at both the base and mobile stations. Simulation results show that the proposed algorithm is superior to other solutions in the literature including the nearest Kronecker product (NKP) algorithm and successive interference cancellation (SIC) method. The proposed algorithm provides a higher data rate compared to other methods. In addition, the proposed design offers higher energy efficiency than the fully-connected architecture.

In 5G networks, the Cloud Radio Access Network (C-RAN) is considered a promising future architecture in terms of minimizing energy consumption and allocating resources efficiently by providing real-time cloud infrastructures, cooperative radio, and centralized data processing. Recently, given their vulnerability to malicious attacks, the security of C-RAN networks has attracted significant attention. Among various anomaly-based intrusion detection techniques, the most promising one is the machine learning-based intrusion detection as it learns without human assistance and adjusts actions accordingly. In this direction, many solutions have been proposed, but they show either low accuracy in terms of attack classification or they offer just a single layer of attack detection. This research focuses on deploying a multi-stage machine learning-based intrusion detection (ML-IDS) in 5G C-RAN that can detect and classify four types of jamming attacks: constant jamming, random jamming, deceptive jamming, and reactive jamming. This deployment enhances security by minimizing the false negatives in C-RAN architectures. The experimental evaluation of the proposed solution is carried out using WSN-DS (Wireless Sensor Networks DataSet), which is a dedicated wireless dataset for intrusion detection. The final classification accuracy of attacks is 94.51% with a 7.84% false negative rate.

Several methods have recently been proposed to analyze speech and automatically infer the personality of the speaker. These methods often rely on prosodic and other hand crafted speech processing features extracted with off-the-shelf toolboxes. To achieve high accuracy, numerous features are typically extracted using complex and highly parameterized algorithms. In this paper, a new method based on feature learning and spectrogram analysis is proposed to simplify the feature extraction process while maintaining a high level of accuracy. The proposed method learns a dictionary of discriminant features from patches extracted in the spectrogram representations of training speech segments. Each speech segment is then encoded using the dictionary, and the resulting feature set is used to perform classification of personality traits. Experiments indicate that the proposed method achieves state-of-the-art results with an important reduction in complexity when compared to the most recent reference methods. The number of features, and difficulties linked to the feature extraction process are greatly reduced as only one type of descriptors is used, for which the 7 parameters can be tuned automatically. In contrast, the simplest reference method uses 4 types of descriptors to which 6 functionals are applied, resulting in over 20 parameters to be tuned.

Smart homes play a crucial role in reducing the residential sector electricity consumption and Greenhouse Gases (GHG) emissions. In this work, we present a time series approach to predict GHG emissions to be integrated into smart home management systems. More specifically, we used Long Short-Term Memory (LSTM), a variant of Recurrent Neural Networks. The prediction results get mean absolute percentage error (MAPE) close to 2 percent when the region under study has an energy matrix mostly based on fossil fuels, less intermittent. For regions in which more renewable sources are present, the MAPE is around 12 percent. However, in either case, LSTM can predict the hours well with smaller emissions among the next 24 hours. Such day-ahead information brings awareness to the users and allows the scheduling of appliances to work in the hours in which the emissions are minimal, reducing them without significantly affecting the consumers' behavior.

A growing number of applications, e.g., video surveillance and medical image analysis, require training recognition systems from large amounts of weakly annotated data, while some targeted interactions with a domain expert are allowed to improve the training process. In such cases, active learning (AL) can reduce labeling costs for training a classifier by querying the expert to provide the labels of most informative instances. This paper focuses on AL methods for instance classification problems in multiple instance learning (MIL), where data are arranged into sets, called bags, which are weakly labeled. Most AL methods focus on single-instance learning problems. These methods are not suitable for MIL problems because they cannot account for the bag structure of data. In this paper, new methods for bag-level aggregation of instance informativeness are proposed for multiple instance AL (MIAL). The aggregated informativeness method identifies the most informative instances based on classifier uncertainty and queries bags incorporating the most information. The other proposed method, called cluster-based aggregative sampling, clusters data hierarchically in the instance space. The informativeness of instances is assessed by considering bag labels, inferred instance labels, and the proportion of labels that remain to be discovered in clusters. Both proposed methods significantly outperform reference methods in extensive experiments using benchmark data from several application domains. Results indicate that using an appropriate strategy to address MIAL problems yields a significant reduction in the number of queries needed to achieve the same level of performance as single-instance AL methods.

In this paper, a novel programmable time-difference amplifier (TDA) with femtosecond resolution is presented. The proposed TDA design cascades three time latches and digital gates to amplify the input time difference with selectable gains. The time gain is simply determined by changing digital switches performing time addition. Based on digital gates, the proposed circuit achieves performance comparable to the state-of-the-art TDAs, without the limitations of analog components imposed by scaled CMOS processes. The proposed TDA is simulated in a 1V 65 nm TSMC CMOS process to validate the accuracy of the proposed architecture. The linearity of the proposed TDA is verified from 150 fs to 200 ps with a gain error of less than ±4%, while consuming 519 μW with a 450 MHz clock frequency.

This paper derives the equations leading to the analytic expression of the frequency spectrum at the output of multi-bit delta-sigma modulators afflicted by component mismatch in the digital-to-analog converter used in the feedback path. The effect of the mismatch is modeled as an error signal added to an ideal digital-to-analog converter. The frequency content of this error signal is derived from the probability density functions of the input signal and the shaped quantization noise. The analysis is applied to band-limited Gaussian input signals. Several simulation results are reported, showing a 0.3 dB accuracy of the analytic expressions whenever the number of quantization bits is higher than two.

Multiple instance learning (MIL) is a form of weakly-supervised learning where instances are organized in bags. A label is provided for bags, but not for instances. MIL literature typically focuses on the classification of bags seen as one object, or as a combination of their instances. In both cases, performance is generally measured using labels assigned to entire bags. In this paper, the MIL problem is formulated as a knowledge discovery task for which algorithms seek to discover the witnesses (i.e. identifying positive instances), using the weak supervision provided by bag labels. Some MIL methods are suitable for instance classification, but perform poorly in application where the witness rate is low, or when the positive class distribution is multimodal. A new method that clusters data projected in random subspaces is proposed to perform witness identification in these adverse settings. The proposed method is assessed on MIL data sets from three application domains, and compared to 7 reference MIL algorithms for the witness identification task. The proposed algorithm constantly ranks among the best methods in all experiments, while all other methods perform unevenly across data sets.

This paper presents wireless sensor network unmanned aerial vehicle (WSN-UAV) system for military remote monitoring and surveillance. Large scale WSN is deployed in a battlefield or wide hostile region to collect information of interest and send it to a UAV. Collaborative beamforming (CB) is used to achieve the ground-to-air transmissions. An automatic dependent surveillance-broadcast (ADS-B) based multilateration is used to obtain the UAV location and tracking information. It is found that a minimum distance between the UAV and the WSN is required for proper operation of the CB due to the precision of the multilateration and the movement of the UAV.

A cognitive detect and avoid radar system based on chaotic UWB-MIMO waveform design to enable autonomous UAV navigation is presented. A Dirichlet-process-mixture-model (DPMM)-based Bayesian clustering approach to discriminate extended targets and a change-point (CP) detection algorithm are applied for the autonomous tracking and identification of potential collision threats. A DPMM-based clustering mechanism does not rely upon any a priori target scene assumptions and facilitates online multivariate data clustering/classification for an arbitrary number of targets. Furthermore, this radar system utilizes a cognitive mechanism to select efficient chaotic waveforms to facilitate enhanced target detection and discrimination. We formulate the CP mechanism for the online tracking of target trajectories, which present a collision threat to the UAV navigation; thus, we supplement the conventional Kalman-filter-based tracking. Simulation results demonstrate a significant performance improvement for the DPMM-CP-assisted detection as compared with direct generalized likelihood-ratio-based detection. Specifically, we observe a 4-dB performance gain in target detection over conventional fixed UWB waveforms and superior collision avoidance capability offered by the joint DPMM-CP mechanism.

The detection of play and break segments in team sports is an essential step towards the automation of live game capture and broadcast. This paper presents a two-stage hierarchical method for play-break detection in non-edited video feeds of sport events. Unlike most existing methods, this algorithm performs action and event recognition on content, and thus does not rely on production cues of broadcast feeds. Moreover, the method does not require player tracking, can be used in real-time, and can be easily adapted to different sports. In the first stage, bag-of-words event detectors are trained to recognize key events such as line changes, face-offs and preliminary play-breaks. In the second stage, the output of the detectors along with a novel feature based on spatio-temporal interest points are used to create a context descriptor for the final decision. Experiments demonstrate the efficiency of the proposed method on real hockey game footage, achieving 90% accuracy.

A voltage-controlled delay unit (VCDU) for low-voltage time-mode signal processing is presented in this paper. The proposed VCDU uses a signal conditioning circuit to achieve wider-range and higher linearity than state-of-the-art VCDUs. Circuit-level simulations in 0.18μm CMOS process show a linearity error of less than ± 0.2% for a 0.15 V to 1 V input range. The proposed VCDU consumes 315 μW from a 1.8 V supply at its maximum sampling frequency of 500 MHz. The proposed VCDU is validated in a first-order time-mode ΔΣ modulator application. Circuit-level simulation results of the ΔΣ modulator show a peak SNDR of 58 dB when clocked at 140 MHz with a 400 kHz bandwidth.

In this paper, a Voice Activity Detector (VAD) is proposed for smart hearing protection applications where speech is to get through the hearing protector while ambient noise is to be blocked out.The VAD calculates a short-term statistical assessment of the temporal envelopes within different frequency bands.This assessment uses the Inter-Quartile Range (IQR) and reflects the dispersion of the envelopes' magnitudes.The VAD's decision is made using two threshold comparison rules and a hangover scheme triggered after a given number of observations.These four parameters have been optimized off-line using a genetic algorithm approach.The performance of the proposed VAD is compared to Sohn's VAD using a database of 90 speech signals corrupted by five real-world noise environments at Signal-to-Noise ratios (SNR) varying from 0 to +10 dB.Results show that the proposed VAD performs better than Sohn's VAD with an 85.9% (compared to 77.5%) F1 score averaged across all SNRs and also minimizes by a factor of three the mid-speech clipping rate.In addition, the evaluation of the proposed VAD's computational cost shows that its implementation on-board a low-power low-consumption DSP is very feasible and would enable smart hearing protection for hypersensitive persons.

In this paper, we propose a robust synchronization technique for an asynchronous spread spectrum communication system based on chaotic symbolic dynamics modulation. A back-ward iteration of the chaotic map is used to avoid the problem of sensitivity to initial conditions of the chaotic generator. The proposed system integrates a spread spectrum unit, which is used to increase transmission security and to achieve the transmission in a multi-user case. The synchronization technique is assessed in terms of probability of detection and of probability of false alarm. Simulation results prove that the proposed system can achieve phase synchronization with a low signal-to-noise ratio.

We present a novel calibration technique to compensate for DAC element mismatches in bandpass multibit delta-sigma (Δ-Σ) modulators. The proposed technique is purely digital and requires only a minor modification to the modulator loop. It is compatible with binary weighted element DACs and the storage requirements for the calibrated coefficients increases only linearly with the number of quantizer bits. The calibration is performed without breaking the loop, which allows continuous tracking of environmental drifts. Simulation results show a peak signal to noise and distortion ratio (SNDR) of 68 dB after calibration for a DAC with±1%mismatches, a sinusoid input signal near 1/4 of the sampling frequency and an oversampling ratio of only 10. Those results represent a 26 dB improvement over the non-calibrated case while being within 2 dB of an ideal-DAC case.