Michele Rossi

Francesca Meneghello, Michele Rossi, Francesco Restuccia

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.

Seyyidahmed Lahmer, Aria Khoshsirat, Michele Rossi et al.

Recently, there has been a trend of shifting the execution of deep learning inference tasks toward the edge of the network, closer to the user, to reduce latency and preserve data privacy. At the same time, growing interest is being devoted to the energetic sustainability of machine learning. At the intersection of these trends, we hence find the energetic characterization of machine learning at the edge, which is attracting increasing attention. Unfortunately, calculating the energy consumption of a given neural network during inference is complicated by the heterogeneity of the possible underlying hardware implementation. In this work, we hence aim at profiling the energetic consumption of inference tasks for some modern edge nodes and deriving simple but realistic models. To this end, we performed a large number of experiments to collect the energy consumption of convolutional and fully connected layers on two well-known edge boards by NVIDIA, namely Jetson TX2 and Xavier. From the measurements, we have then distilled a simple, practical model that can provide an estimate of the energy consumption of a certain inference task on the considered boards. We believe that this model can be used in many contexts as, for instance, to guide the search for efficient architectures in Neural Architecture Search, as a heuristic in Neural Network pruning, or to find energy-efficient offloading strategies in a Split computing context, or simply to evaluate the energetic performance of Deep Neural Network architectures.

Francesca Meneghello, Nicolò Dal Fabbro, Domenico Garlisi et al.

In the last years, several machine learning-based techniques have been proposed to monitor human movements from Wi-Fi channel readings. However, the development of domain-adaptive algorithms that robustly work across different environments is still an open problem, whose solution requires large datasets characterized by strong domain diversity, in terms of environments, persons and Wi-Fi hardware. To date, the few public datasets available are mostly obsolete - as obtained via Wi-Fi devices operating on 20 or 40 MHz bands - and contain little or no domain diversity, thus dramatically limiting the advancements in the design of sensing algorithms. The present contribution aims to fill this gap by providing a dataset of IEEE 802.11ac channel measurements over an 80 MHz bandwidth channel featuring notable domain diversity, through measurement campaigns that involved thirteen subjects across different environments, days, and with different hardware. Novel experimental data is provided by blocking the direct path between the transmitter and the monitor, and collecting measurements in a semi-anechoic chamber (no multi-path fading). Overall, the dataset - available on IEEE DataPort [1] - contains more than thirteen hours of channel state information readings (23.6 GB), allowing researchers to test activity/identity recognition and people counting algorithms.

Luca Ballotta, Nicolò Dal Fabbro, Giovanni Perin et al.

Assisted and autonomous driving are rapidly gaining momentum and will soon become a reality. Artificial intelligence and machine learning are regarded as key enablers thanks to the massive amount of data that smart vehicles will collect from onboard sensors. Federated learning is one of the most promising techniques for training global machine learning models while preserving data privacy of vehicles and optimizing communications resource usage. In this article, we propose vehicular radio environment map federated learning (VREM-FL), a computation-scheduling co-design for vehicular federated learning that combines mobility of vehicles with 5G radio environment maps. VREM-FL jointly optimizes learning performance of the global model and wisely allocates communication and computation resources. This is achieved by orchestrating local computations at the vehicles in conjunction with transmission of their local models in an adaptive and predictive fashion, by exploiting radio channel maps. The proposed algorithm can be tuned to trade training time for radio resource usage. Experimental results demonstrate that VREM-FL outperforms literature benchmarks for both a linear regression model (learning time reduced by 28%) and a deep neural network for semantic image segmentation (doubling the number of model updates within the same time window).

Lúcio Folly Sanches Zebendo, Eleonora Cicciarella, Michele Rossi

Spiking neural networks (SNNs) are rapidly gaining momentum as an alternative to conventional artificial neural networks in resource constrained edge systems. In this work, we continue a recent research line on recurrent SNNs where axonal delays are learned at runtime along with the other network parameters. The first proposed approach, dubbed DelRec, demonstrated the benefit of recurrent delay learning in SNNs. Here, we extend it by advocating the use of convolutional recurrent connections in conjunction with the DelRec delay learning mechanism. According to our tests on an audio classification task, this leads to a streamlined architecture with smaller memory footprint (around 99% savings in terms of number of recurrent parameters) and a much faster (52x) inference time, while retaining DelRec's accuracy. Our code is available at: https://github.com/luciozebendo/delrec_snn/tree/conv_delays

Riccardo Mazzieri, Jacopo Pegoraro, Michele Rossi

The reconstruction of micro-Doppler signatures of human movements is a key enabler for fine-grained activity recognition wireless sensing. In Joint Communication and Sensing (JCS) systems, unlike in dedicated radar sensing systems, a suitable trade-off between sensing accuracy and communication overhead has to be attained. It follows that the micro-Doppler has to be reconstructed from incomplete windows of channel estimates obtained from communication packets. Existing approaches exploit compressed sensing, but produce very poor reconstructions when only a few channel measurements are available, which is often the case with real communication patterns. In addition, the large number of iterations they need to converge hinders their use in real-time systems. In this work, we propose and validate STAR, a neural network that reconstructs micro-Doppler sequences of human movement even from highly incomplete channel measurements. STAR is based upon a new architectural design that combines a single unrolled iterative hard-thresholding layer with an attention mechanism, used at its output. This results in an interpretable and lightweight architecture that reaps the benefits of both model-based and data driven solutions. STAR is evaluated on a public JCS dataset of 60 GHz channel measurements of human activity traces. Experimental results show that it substantially outperforms state-of-the-art techniques in terms of the reconstructed micro-Doppler quality. Remarkably, STAR enables human activity recognition with satisfactory accuracy even with 90% of missing channel measurements, for which existing techniques fail.

Zaman Bhalli, Michele Rossi, Joerg Widmer et al.

Integrated Sensing And Communication (ISAC) is recognized as a key enabler for future 6th Generation (6G) networks, combining communication capabilities with pervasive sensing. In such systems, the estimation of the Doppler shift plays a crucial role for target characterization. However, typical real-world ISAC scenarios largely involve bistatic or multistatic configurations and mobile ISAC nodes. Under these conditions, Doppler estimation becomes particularly challenging, as clock asynchrony between the Transmitter (TX) and the Receivers (RXs), combined with their mobility, introduces additional Doppler components and phase offsets that distort or disrupt the target-induced frequency shift. Existing works have considered these challenges separately or relied on external reference reflectors. In this paper, we present the first method to estimate the Doppler frequency of a target with mobile and asynchronous ISAC nodes in a multistatic configuration, considering the case of a mobile TX and multiple static RXs, and without leveraging any external reflector. By leveraging the invariance of the phase offsets across multipath components and exploiting geometrical relationships, we show that the problem is solvable if at least 4 RXs are present. We evaluate the proposed solution through numerical simulations in various scenarios, showing that it is a valid approach for estimating target Doppler shifts in unsynchronized multistatic ISAC deployments with mobile nodes.

Yesmine Abdennadher, Eleonora Cicciarella, Michele Rossi

Spike-based temporal messaging enables SNNs to efficiently process both purely temporal and spatio-temporal time-series or event-driven data. Combining SNNs with Gated Recurrent Units (GRUs), a variant of recurrent neural networks, gives rise to a robust framework for sequential data processing; however, traditional RNNs often lose local details when handling long sequences. Previous approaches, such as SpikGRU, fail to capture fine-grained local dependencies in event-based spatio-temporal data. In this paper, we introduce the Convolutional Spiking GRU (CS-GRU) cell, which leverages convolutional operations to preserve local structure and dependencies while integrating the temporal precision of spiking neurons with the efficient gating mechanisms of GRUs. This versatile architecture excels on both temporal datasets (NTIDIGITS, SHD) and spatio-temporal benchmarks (MNIST, DVSGesture, CIFAR10DVS). Our experiments show that CS-GRU outperforms state-of-the-art GRU variants by an average of 4.35%, achieving over 90% accuracy on sequential tasks and up to 99.31% on MNIST. It is worth noting that our solution achieves 69% higher efficiency compared to SpikGRU. The code is available at: https://github.com/YesmineAbdennadher/CS-GRU.

Yesmine Abdennadher, Giovanni Perin, Riccardo Mazzieri et al.

Spiking Neural Networks (SNNs) are highly regarded for their energy efficiency, inherent activation sparsity, and suitability for real-time processing in edge devices. However, most current SNN methods adopt architectures resembling traditional artificial neural networks (ANNs), leading to suboptimal performance when applied to SNNs. While SNNs excel in energy efficiency, they have been associated with lower accuracy levels than traditional ANNs when utilizing conventional architectures. In response, in this work we present LightSNN, a rapid and efficient Neural Network Architecture Search (NAS) technique specifically tailored for SNNs that autonomously leverages the most suitable architecture, striking a good balance between accuracy and efficiency by enforcing sparsity. Based on the spiking NAS network (SNASNet) framework, a cell-based search space including backward connections is utilized to build our training-free pruning-based NAS mechanism. Our technique assesses diverse spike activation patterns across different data samples using a sparsity-aware Hamming distance fitness evaluation. Thorough experiments are conducted on both static (CIFAR10 and CIFAR100) and neuromorphic datasets (DVS128-Gesture). Our LightSNN model achieves state-of-the-art results on CIFAR10 and CIFAR100, improves performance on DVS128Gesture by 4.49\%, and significantly reduces search time most notably offering a $98\times$ speedup over SNASNet and running 30\% faster than the best existing method on DVS128Gesture. Code is available on Github at: https://github.com/YesmineAbdennadher/LightSNN.

Riccardo Mazzieri, Jacopo Pegoraro, Michele Rossi

The adoption of Millimeter-Wave (mmWave) radar devices for human sensing, particularly gait recognition, has recently gathered significant attention due to their efficiency, resilience to environmental conditions, and privacy-preserving nature. In this work, we tackle the challenging problem of Open-set Gait Recognition (OSGR) from sparse mmWave radar point clouds. Unlike most existing research, which assumes a closed-set scenario, our work considers the more realistic open-set case, where unknown subjects might be present at inference time, and should be correctly recognized by the system. Point clouds are well-suited for edge computing applications with resource constraints, but are more significantly affected by noise and random fluctuations than other representations, like the more common micro-Doppler signature. This is the first work addressing open-set gait recognition with sparse point cloud data. To do so, we propose a novel neural network architecture that combines supervised classification with unsupervised reconstruction of the point clouds, creating a robust, rich, and highly regularized latent space of gait features. To detect unknown subjects at inference time, we introduce a probabilistic novelty detection algorithm that leverages the structured latent space and offers a tunable trade-off between inference speed and prediction accuracy. Along with this paper, we release mmGait10, an original human gait dataset featuring over five hours of measurements from ten subjects, under varied walking modalities. Extensive experimental results show that our solution attains F1-Score improvements by 24% over state-of-the-art methods adapted for point clouds, on average, and across multiple openness levels.

Giovanni Perin, Cesare Bidini, Riccardo Mazzieri et al.

In recent years, spiking neural networks (SNNs) have gained momentum due to their high potential in time-series processing combined with minimal energy consumption. However, they still lack a dedicated and efficient training algorithm. The popular backpropagation with surrogate gradients, adapted from stochastic gradient descent (SGD)-derived algorithms, has several drawbacks when used as an optimizer for SNNs. Specifically, it suffers from low scalability and numerical imprecision. In this paper, we propose a novel SNN training method based on the alternating direction method of multipliers (ADMM). Our ADMM-based training aims to solve the problem of the SNN step function's non-differentiability. We formulate the problem, derive closed-form updates, and empirically show the optimizer's convergence properties, great potential, and possible new research directions to improve the method in a simulated proof-of-concept.

Nicolò Dal Fabbro, Michele Rossi, Luca Schenato et al.

Edge networks call for communication efficient (low overhead) and robust distributed optimization (DO) algorithms. These are, in fact, desirable qualities for DO frameworks, such as federated edge learning techniques, in the presence of data and system heterogeneity, and in scenarios where internode communication is the main bottleneck. Although computationally demanding, Newton-type (NT) methods have been recently advocated as enablers of robust convergence rates in challenging DO problems where edge devices have sufficient computational power. Along these lines, in this work we propose Q-SHED, an original NT algorithm for DO featuring a novel bit-allocation scheme based on incremental Hessian eigenvectors quantization. The proposed technique is integrated with the recent SHED algorithm, from which it inherits appealing features like the small number of required Hessian computations, while being bandwidth-versatile at a bit-resolution level. Our empirical evaluation against competing approaches shows that Q-SHED can reduce by up to 60% the number of communication rounds required for convergence.

Nicolò Dal Fabbro, Subhrakanti Dey, Michele Rossi et al.

There is a growing interest in the distributed optimization framework that goes under the name of Federated Learning (FL). In particular, much attention is being turned to FL scenarios where the network is strongly heterogeneous in terms of communication resources (e.g., bandwidth) and data distribution. In these cases, communication between local machines (agents) and the central server (Master) is a main consideration. In this work, we present SHED, an original communication-constrained Newton-type (NT) algorithm designed to accelerate FL in such heterogeneous scenarios. SHED is by design robust to non i.i.d. data distributions, handles heterogeneity of agents' communication resources (CRs), only requires sporadic Hessian computations, and achieves super-linear convergence. This is possible thanks to an incremental strategy, based on eigendecomposition of the local Hessian matrices, which exploits (possibly) outdated second-order information. The proposed solution is thoroughly validated on real datasets by assessing (i) the number of communication rounds required for convergence, (ii) the overall amount of data transmitted and (iii) the number of local Hessian computations. For all these metrics, the proposed approach shows superior performance against state-of-the art techniques like GIANT and FedNL.

Anish Shastri, Neharika Valecha, Enver Bashirov et al.

The commercial availability of low-cost millimeter wave (mmWave) communication and radar devices is starting to improve the penetration of such technologies in consumer markets, paving the way for large-scale and dense deployments in fifth-generation (5G)-and-beyond as well as 6G networks. At the same time, pervasive mmWave access will enable device localization and device-free sensing with unprecedented accuracy, especially with respect to sub-6 GHz commercial-grade devices. This paper surveys the state of the art in device-based localization and device-free sensing using mmWave communication and radar devices, with a focus on indoor deployments. We first overview key concepts about mmWave signal propagation and system design. Then, we provide a detailed account of approaches and algorithms for localization and sensing enabled by mmWaves. We consider several dimensions in our analysis, including the main objectives, techniques, and performance of each work, whether each research reached some degree of implementation, and which hardware platforms were used for this purpose. We conclude by discussing that better algorithms for consumer-grade devices, data fusion methods for dense deployments, as well as an educated application of machine learning methods are promising, relevant and timely research directions.

Marco Canil, Jacopo Pegoraro, Michele Rossi

Social distancing and temperature screening have been widely employed to counteract the COVID-19 pandemic, sparking great interest from academia, industry and public administrations worldwide. While most solutions have dealt with these aspects separately, their combination would greatly benefit the continuous monitoring of public spaces and help trigger effective countermeasures. This work presents milliTRACE-IR, a joint mmWave radar and infrared imaging sensing system performing unobtrusive and privacy preserving human body temperature screening and contact tracing in indoor spaces. milliTRACE-IR combines, via a robust sensor fusion approach, mmWave radars and infrared thermal cameras. It achieves fully automated measurement of distancing and body temperature, by jointly tracking the subjects's faces in the thermal camera image plane and the human motion in the radar reference system. Moreover, milliTRACE-IR performs contact tracing: a person with high body temperature is reliably detected by the thermal camera sensor and subsequently traced across a large indoor area in a non-invasive way by the radars. When entering a new room, a subject is re-identified among several other individuals by computing gait-related features from the radar reflections through a deep neural network and using a weighted extreme learning machine as the final re-identification tool. Experimental results, obtained from a real implementation of milliTRACE-IR, demonstrate decimeter-level accuracy in distance/trajectory estimation, inter-personal distance estimation (effective for subjects getting as close as 0.2 m), and accurate temperature monitoring (max. errors of 0.5°C). Furthermore, milliTRACE-IR provides contact tracing through highly accurate (95%) person re-identification, in less than 20 seconds.

Francesca Meneghello, Domenico Garlisi, Nicolò Dal Fabbro et al.

In this article we present SHARP, an original approach for obtaining human activity recognition (HAR) through the use of commercial IEEE 802.11 (Wi-Fi) devices. SHARP grants the possibility to discern the activities of different persons, across different time-spans and environments. To achieve this, we devise a new technique to clean and process the channel frequency response (CFR) phase of the Wi-Fi channel, obtaining an estimate of the Doppler shift at a radio monitor device. The Doppler shift reveals the presence of moving scatterers in the environment, while not being affected by (environment-specific) static objects. SHARP is trained on data collected as a person performs seven different activities in a single environment. It is then tested on different setups, to assess its performance as the person, the day and/or the environment change with respect to those considered at training time. In the worst-case scenario, it reaches an average accuracy higher than 95%, validating the effectiveness of the extracted Doppler information, used in conjunction with a learning algorithm based on a neural network, in recognizing human activities in a subject and environment independent way. The collected CFR dataset and the code are publicly available for replicability and benchmarking purposes.

Riccardo Bonetto, Mattia Soldan, Alberto Lanaro et al.

Smartphones and wearable devices are fast growing technologies that, in conjunction with advances in wireless sensor hardware, are enabling ubiquitous sensing applications. Wearables are suitable for indoor and outdoor scenarios, can be placed on many parts of the human body and can integrate a large number of sensors capable of gathering physiological and behavioral biometric information. Here, we are concerned with gait analysis systems that extract meaningful information from a user's movements to identify anomalies and changes in their walking style. The solution that is put forward is subject-specific, as the designed feature extraction and classification tools are trained on the subject under observation. A smartphone mounted on an ad-hoc made chest support is utilized to gather inertial data and video signals from its built-in sensors and rear-facing camera. The collected video and inertial data are preprocessed, combined and then classified by means of a Recurrent Neural Network (RNN) based Sequence-to-Sequence (Seq2Seq) model, which is used as a feature extractor, and a following Convolutional Neural Network (CNN) classifier. This architecture provides excellent results, being able to correctly assess anomalies in 100% of the cases, for the considered tests, surpassing the performance of support vector machine classifiers.

Hoang Duy Trinh, Angel Fernandez Gambin, Lorenza Giupponi et al.

The automatic classification of applications and services is an invaluable feature for new generation mobile networks. Here, we propose and validate algorithms to perform this task, at runtime, from the raw physical channel of an operative mobile network, without having to decode and/or decrypt the transmitted flows. Towards this, we decode Downlink Control Information (DCI) messages carried within the LTE Physical Downlink Control CHannel (PDCCH). DCI messages are sent by the radio cell in clear text and, in this paper, are utilized to classify the applications and services executed at the connected mobile terminals. Two datasets are collected through a large measurement campaign: one labeled, used to train the classification algorithms, and one unlabeled, collected from four radio cells in the metropolitan area of Barcelona, in Spain. Among other approaches, our Convolutional Neural Network (CNN) classifier provides the highest classification accuracy of 99%. The CNN classifier is then augmented with the capability of rejecting sessions whose patterns do not conform to those learned during the training phase, and is subsequently utilized to attain a fine grained decomposition of the traffic for the four monitored radio cells, in an online and unsupervised fashion.

Matteo Gadaleta, Giulia Cisotto, Michele Rossi et al.

Wearable technology for the automatic detection of gait events has recently gained growing interest, enabling advanced analyses that were previously limited to specialist centres and equipment (e.g., instrumented walkway). In this study, we present a novel method based on dilated convolutions for an accurate detection of gait events (initial and final foot contacts) from wearable inertial sensors. A rich dataset has been used to validate the method, featuring 71 people with Parkinson's disease (PD) and 67 healthy control subjects. Multiple sensors have been considered, one located on the fifth lumbar vertebrae and two on the ankles. The aims of this study were: (i) to apply deep learning (DL) techniques on wearable sensor data for gait segmentation and quantification in older adults and in people with PD; (ii) to validate the proposed technique for measuring gait against traditional gold standard laboratory reference and a widely used algorithm based on wavelet transforms (WT); (iii) to assess the performance of DL methods in assessing high-level gait characteristics, with focus on stride, stance and swing related features. The results showed a high reliability of the proposed approach, which achieves temporal errors considerably smaller than WT, in particular for the detection of final contacts, with an inter-quartile range below 70 ms in the worst case. This study showes encouraging results, and paves the road for further research, addressing the effectiveness and the generalization of data-driven learning systems for accurate event detection in challenging conditions.

Riccardo Bonetto, Michele Rossi

We consider the problem of power demand forecasting in residential micro-grids. Several approaches using ARMA models, support vector machines, and recurrent neural networks that perform one-step ahead predictions have been proposed in the literature. Here, we extend them to perform multi-step ahead forecasting and we compare their performance. Toward this end, we implement a parallel and efficient training framework, using power demand traces from real deployments to gauge the accuracy of the considered techniques. Our results indicate that machine learning schemes achieve smaller prediction errors in the mean and the variance with respect to ARMA, but there is no clear algorithm of choice among them. Pros and cons of these approaches are discussed and the solution of choice is found to depend on the specific use case requirements. A hybrid approach, that is driven by the prediction interval, the target error, and its uncertainty, is then recommended.

Davide Zordan, Michele Rossi, Michele Zorzi

Many future wireless sensor networks and the Internet of Things are expected to follow a software defined paradigm, where protocol parameters and behaviors will be dynamically tuned as a function of the signal statistics. New protocols will be then injected as a software as certain events occur. For instance, new data compressors could be (re)programmed on-the-fly as the monitored signal type or its statistical properties change. We consider a lossy compression scenario, where the application tolerates some distortion of the gathered signal in return for improved energy efficiency. To reap the full benefits of this paradigm, we discuss an automatic sensor profiling approach where the signal class, and in particular the corresponding rate-distortion curve, is automatically assessed using machine learning tools (namely, support vector machines and neural networks). We show that this curve can be reliably estimated on-the-fly through the computation of a small number (from ten to twenty) of statistical features on time windows of a few hundreds samples.

Riccardo Bonetto, Michele Rossi, Stefano Tomasin et al.

In electrical distribution grids, the constantly increasing number of power generation devices based on renewables demands a transition from a centralized to a distributed generation paradigm. In fact, power injection from Distributed Energy Resources (DERs) can be selectively controlled to achieve other objectives beyond supporting loads, such as the minimization of the power losses along the distribution lines and the subsequent increase of the grid hosting capacity. However, these technical achievements are only possible if alongside electrical optimization schemes, a suitable market model is set up to promote cooperation from the end users. In contrast with the existing literature, where energy trading and electrical optimization of the grid are often treated separately or the trading strategy is tailored to a specific electrical optimization objective, in this work we consider their joint optimization. Specifically, we present a multi-objective optimization problem accounting for energy trading, where: 1) DERs try to maximize their profit, resulting from selling their surplus energy, 2) the loads try to minimize their expense, and 3) the main power supplier aims at maximizing the electrical grid efficiency through a suitable discount policy. This optimization problem is proved to be non convex, and an equivalent convex formulation is derived. Centralized solutions are discussed first, and are subsequently distributed. Numerical results to demonstrate the effectiveness of the so obtained optimal policies are then presented.

Matteo Gadaleta, Michele Rossi

Here, we present IDNet, a user authentication framework from smartphone-acquired motion signals. Its goal is to recognize a target user from their way of walking, using the accelerometer and gyroscope (inertial) signals provided by a commercial smartphone worn in the front pocket of the user's trousers. IDNet features several innovations including: i) a robust and smartphone-orientation-independent walking cycle extraction block, ii) a novel feature extractor based on convolutional neural networks, iii) a one-class support vector machine to classify walking cycles, and the coherent integration of these into iv) a multi-stage authentication technique. IDNet is the first system that exploits a deep learning approach as universal feature extractors for gait recognition, and that combines classification results from subsequent walking cycles into a multi-stage decision making framework. Experimental results show the superiority of our approach against state-of-the-art techniques, leading to misclassification rates (either false negatives or positives) smaller than 0.15% with fewer than five walking cycles. Design choices are discussed and motivated throughout, assessing their impact on the user authentication performance.