Searcharxiv⌕ Search

arXiv subjects

Michele Rossi

Publications and source records attributed to Michele Rossi.

At least 19 recordsLinked to original sources

VISTA: View-Independent micro-Doppler Reconstruction with Multistatic ISAC

The micro-Doppler (uD) spectrum provides detailed characteristics of target movement, making it highly valuable for applications such as target classification and human activity recognition. However, traditional Integrated Sensing And Communication (ISAC) systems are highly view-dependent and limited by blind spots and angle dependencies. Existing approaches rely on either configuration-specific solutions or data-driven models. The former results in poor generalization when device deployments or target orientations change, while the latter requires acquiring extensive multi-view datasets, which is time-consuming and not always feasible. To address these limitations, in this work we propose VISTA, a novel method for reconstructing the uD spectrogram in an absolute coordinate frame, yielding view-independent Doppler features that are invariant to the spatial distribution of ISAC nodes. We formulate the uD estimation as a sparse reconstruction problem that jointly leverages measurements from all ISAC receivers. This is solved using a modified Orthogonal Matching Pursuit (OMP) algorithm, which iteratively alternates between jointly selecting candidate velocity vectors across all receivers (RXs) and independently estimating the complex channel gain amplitudes for each RX. Our results show that, provided sufficient spatial diversity, VISTA successfully reconstructs the targets' absolute uD spectrograms with a median error \leq 0.69 m/s for speeds \leq 15 m/s.

cs.AR↗

Multi-Depth Temporal Fusion for Feedforward, Locally Trained Spiking Neural Networks

We propose a new spiking neural network (SNN) design to process static images and event streams using time-to-first-spike (TTFS) latencies. Our key research question is which architectural choices best accommodate local and online learning in multi-layer convolutional SNNs. This question is addressed via an original framework combining residual-like connections with multi-depth feature aggregation and consensus. The full SNN pipeline features an early-vision front end, to convert raw visual data into sparse spike latencies, a four-layer convolutional backbone trained layerwise with unsupervised spike-timing-dependent plasticity (STDP), a deterministic Multi-Depth Temporal Fusion (MDTF) and a final classifier trained with reward-modulated spike-timing-dependent plasticity (R-STDP). Rather than replacing early features in deeper layers, the proposed MDTF preserves early temporal evidence, adding sparse residual events from intermediate layers, and incorporating deeper features only when they agree in time with earlier representations. The resulting architecture is experimentally validated across MNIST, Fashion-MNIST, CIFAR-10, and N-MNIST, delivering strong classification performance under a fully local learning regime. Selective multi-depth fusion significantly outperforms traditional STDP/R-STDP baselines on higher-variability visual tasks (achieving +18.2 pp on Fashion-MNIST and +29.2 pp on CIFAR-10). Furthermore, activity-budget analyses show that the network retains high accuracy even when removing a large fraction of late or weak spike events, confirming its high data efficiency and reduced event-processing requirements. The codebase is publicly available at github.com/aidinattar/multi-depth-temporal-fusion-snn.

cs.NE↗

Reliable Doppler Estimation in Asynchronous Moving ISAC Devices via IMU Integration

We present an Integrated Sensing And Communication (ISAC) framework for the joint estimation of the Doppler frequency of a passive mobile target in a bistatic scenario with clock-asynchronous nodes and where the Receiver (RX) is static but the Transmitter (TX) is mobile. In such a setup, previous geometric solutions are integrated with an Inertial Measurement Unit (IMU) device at the TX, coming up with a truly joint data fusion and estimation algorithm based on extended Kalman filtering. The approach jointly estimates the target Doppler frequency, along with the speed and direction of motion of the TX and it is robust to the unavailability of static paths between the TX and RX pair, a condition that makes previous solutions ineffective. The developed extended Kalman filter strikes a balance between the accuracy of ISAC and the IMU reliability. The proposed solution is validated via numerical simulations, obtaining a median Doppler error of 1.2% with a smartphone-grade IMU under realistic operating conditions.

eess.SP↗

The multiplicity of a Mori Dream Space: applications to polar duality and the classification of toric $\mathbb{Q}$-Fano varieties

We extend the multiplicity of fake weighted projective spaces, as introduced by Averkov, Kasprzyk, Lehmann and Nill in 2021, to complete toric varieties and Mori Dream Spaces, defining it as the order of the torsion subgroup of the divisor class group. It equals the degree of the canonical $1$-covering, a finite surjective morphism étale in codimension one. We classify toric $\Q$-Fano varieties with prescribed weight matrix and Gorenstein index by admissible subgroups of a finite abelian group, realized as the deck transformation groups of their universal $1$-coverings. The classification allows arbitrary class-group rank without assuming $\Q$-factoriality, and yields an intrinsic intermediate $1$-covering, the \emph{unitary} one, controlling the Gorenstein index. We outline a computational classification strategy based on weight matrices and finite subgroup enumeration. In the Gorenstein Fano case, a perfect pairing of weight groups expresses polar duality as an inclusion-reversing subgroup correspondence. For iso-weighted polar partners over $\C$, the mirror correspondence admits an internal description within the anticanonical family of their common universal $1$-covering, through invariant subfamilies and their quotient actions. Finally, we derive divisibility relations and multiplicity bounds, also applicable to Mori Dream Spaces whose canonical toric ambient varieties admit suitable $\Q$-Fano completions.

math.AG↗

Predictive Suppression Layers for Communication-Efficient Spiking Neural Networks

Feedforward Spiking Neural Networks (SNNs) typically propagate every generated spike indiscriminately, disregarding whether the information is redundant from an information-theoretic perspective. This lack of selectivity induces high redundancy in inter-layer communication, creating an expensive overhead, e.g., in scenarios involving many-core neuromorphic hardware or communication-dominated Internet-of-Things (IoT) where features are transmitted wirelessly. To address this challenge, we trade localized processing for leaner network channels by introducing a minimal predictive coding framework for SNNs. We propose two layer variants sharing a predictor block: error units, which transmit signed spiking residuals, and predictive suppression, which uses residual magnitude to dynamically gate and forward only unpredictable, "surprising" activity. Evaluated on the N-MNIST and Spiking Heidelberg Digits (SHD) datasets using diagnostic metrics that decouple local processing from cross-layer communication, our new predictive coding layers achieve significant communication savings. Numerical results reveal a three-fold reduction in communicated activity, while increasing the task accuracy for both datasets. The latter finding is notable, and suggests that predictive coding layers not only minimize communication overhead, but also produce output feature vectors with a higher representation power.

cs.NE↗

Trustworthy, Explainable, and Sustainable Decentralized Intelligence for 6G Networks

As 6G networks transition from theoretical frameworks into operational realities, artificial intelligence (AI) evolves from an add-on optimization tool into a distributed and interconnected structural layer. Unlike previous network generations that mostly relied on centralized cloud analytics platforms, AI-native 6G networks operate across a dynamic, multi-domain edge-cloud continuum where data originates from heterogeneous sources including user devices, radio access networks, sensing infrastructures, and vertical applications. Centralizing this massive volume of data creates severe communication overhead, unacceptable latency bottlenecks, single points of failure, and complex cross-domain governance challenges. Consequently, decentralization becomes a fundamental architectural requirement for future 6G network intelligence and zero-touch operations. Security serves as the primary enabler of this decentralized paradigm. Critical security functions, such as real-time threat detection, physical-layer attack mitigation, slice protection, and intrusion detection, require immediate access to local context and telemetry before operational data loses its value. However, moving intelligence to the edge via collaborative paradigms like federated learning (FL) and decentralized FL (DFL) introduces complex trade-offs. System security cannot be addressed in isolation; it is deeply intertwined with equally important aspects like trustworthiness, explainability, and energy sustainability. Taking these aspects into account, this paper develops a unified perspective on decentralized intelligence for 6G, arguing that decentralization, trustworthiness, explainability, and sustainability must be designed jointly rather than treated as independent requirements

cs.NI↗

Estimating Target Doppler in Unsynchronized Multistatic ISAC Deployments with Mobile Nodes

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.

eess.SP↗

Non-calibrated framed processes, derived equivalence and Homological Mirror Symmetry

The present paper is aimed to discussing three kinds of problems: (1) producing some ``mirror theorem'' for the recent mirror symmetric construction, called \emph{framed} duality ($f$-duality), described in \cite{R-fTV} and \cite{R-fpCI}: this is performed from the point of view proposed by Homological Mirror Symmetry (HMS), by studying \emph{derived equivalence} ($D$-equivalence) of multiple mirror models produced by means of a, so-called, \emph{uncalibrated $f$-process}; (2) proposing a general construction giving a big number of multiple mirror models to, in principle, any projective complete intersection of non-negative Kodaira dimension: these multiple mirrors turn out to be each other connected by means of uncalibrated $f$-processes and then, after (1), $D$-equivalent or $K$-equivalent, in the sense of Kawamata \cite{Kawamata}; (3) presenting a number of evidences for the Bondal-Orlov-Kawamata conjecture that $D$-equivalence is $K$-equivalence, and viceversa.

math.AG↗

Combining Convolution and Delay Learning in Recurrent Spiking Neural Networks

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

cs.NE↗

Zero-Overhead Unambiguous Velocity Estimation in Multiband ISAC Systems Under Random Traffic

This paper proposes an original method for estimating the velocity of a target by leveraging the multiband capabilities of modern Integrated Sensing And Communication (ISAC) systems. Traditional Doppler estimation relies on regular sampling rates, but ISAC systems often face irregular packet arrival times because they reuse opportunistic communication traffic. This non-deterministic timing increases the risk of Doppler ambiguity and aliasing, degrading velocity estimation accuracy. To resolve this, we advocate exploiting frequency diversity across multiple carrier frequencies to observe Doppler shifts without imposing restrictions on packet timing or requiring dedicated sensing overhead. A multiband velocity estimation problem is here formulated as a mixed-integer quadratic program by utilizing phase differences from all possible pairwise packet combinations. By integrating at least one unambiguous phase measurement, the system can reconstruct the true target velocity even under sporadic traffic conditions. Simulation results using realistic traffic traces demonstrate that this approach significantly outperforms multiband likelihood-based and single-band algorithms, with accuracy improving as frequency separation between bands and inter-packet time intervals increase. This framework provides a zero-overhead solution for robust velocity estimation in dynamic ISAC environments.

eess.SP↗

On a conjecture of Fujino and Sato

We revisit results of Fujino--Sato on complete non-projective $\mathbb Q$-factorial toric varieties and their conjectural factorization by flips. We show that their main results admit short conceptual proofs, avoiding any restriction on the dimension and the Picard number, from the general theory of Cox rings and Mori Dream Spaces, once one organizes small $\mathbb Q$-factorial modifications via the GKZ (secondary fan) decomposition of the moving cone. Moreover, we extend this viewpoint beyond the toric case by proving an analogous statement for complete $\mathbb Q$-factorial weak Mori Dream Spaces: any non-projective such variety admits a divisor $D$ and a $D$-flip to a (projective) Mori Dream Space. Our approach highlights the role of chambers and wall-crossing in the secondary fan as a unifying framework for these constructions.

math.AG↗

Sparse Spike Encoding of Channel Responses for Energy Efficient Human Activity Recognition

ISAC enables pervasive monitoring, but modern sensing algorithms are often too complex for energy-constrained edge devices. This motivates the development of learning techniques that balance accuracy performance and energy efficiency. Spiking Neural Networks (SNNs) are a promising alternative, processing information as sparse binary spike trains and potentially reducing energy consumption by orders of magnitude. In this work, we propose a spiking convolutional autoencoder (SCAE) that learns tailored spike-encoded representations of channel impulse responses (CIR), jointly trained with an SNN for human activity recognition (HAR), thereby eliminating the need for Doppler domain preprocessing. The results show that our SCAE-SNN achieves F1 scores comparable to a hybrid approach (almost 96%), while producing substantially sparser spike encoding (81.1% sparsity). We also show that encoding CIR data prior to classification improves both HAR accuracy and efficiency. The code is available at https://github.com/ele-ciccia/SCAE-SNN-HAR.

cs.NE↗

ADMM-Based Training for Spiking Neural Networks

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, the approximation introduced by the use of surrogate gradients leads to numerical imprecision, poor tracking of SNN firing times at training time, and, in turn, poor scalability. 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 by taking an entirely new approach with respect to gradient backpropagation. For the first time, we formulate the SNN training problem as an ADMM-based iterative optimization, derive closed-form updates, and empirically show the optimizer's convergence, its great potential, and discuss future and promising research directions to improve the method to different layer types and deeper architectures.

cs.LG↗

Machine Learning on the Edge for Sustainable IoT Networks: A Systematic Literature Review

The Internet of Things (IoT) has become integral to modern technology, enhancing daily life and industrial processes through seamless connectivity. However, the rapid expansion of IoT systems presents significant sustainability challenges, such as high energy consumption and inefficient resource management. Addressing these issues is critical for the long-term viability of IoT networks. Machine learning (ML), with its proven success across various domains, offers promising solutions for optimizing IoT operations. ML algorithms can learn directly from raw data, uncovering hidden patterns and optimizing processes in dynamic environments. Executing ML at the edge of IoT networks can further enhance sustainability by reducing bandwidth usage, enabling real-time decision-making, and improving data privacy. Additionally, testing ML models on actual hardware is essential to ensure satisfactory performance under real-world conditions, as it captures the complexities and constraints of real-world IoT deployments. Combining ML at the edge and actual hardware testing, therefore, increases the reliability of ML models to effectively improve the sustainability of IoT systems. The present systematic literature review explores how ML can be utilized to enhance the sustainability of IoT networks, examining current methodologies, benefits, challenges, and future opportunities. Through our analysis, we aim to provide insights that will drive future innovations in making IoT networks more sustainable.

eess.SY↗

AsyMov: Integrated Sensing and Communications with Asynchronous Moving Devices

Estimating the Doppler frequency shift caused by moving targets is one of the key objectives of Integrated Sensing And Communication (ISAC) systems, as it enables applications such as target classification, human activity recognition, and gait analysis. In practical scenarios, Doppler estimation is hindered by the movement of transmitter and receiver devices, and by the phase offsets caused by their clock asynchrony. Existing approaches have separately addressed these two aspects, either assuming clock-synchronous moving devices or asynchronous static ones. In fact, jointly tackling device motion and clock asynchrony is extremely challenging, as the Doppler shift from device movement differs for each propagation path and the phase offsets are time-varying. In this work, we present AsyMov, a method to estimate the bistatic Doppler frequency of a target and its velocity in ISAC setups featuring mobile and asynchronous devices. It leverages the channel impulse response at the receiver, by originally exploiting the invariance of phase offsets across propagation paths and the bistatic geometry, where the target Doppler and the device velocity are jointly estimated by a newly proposed alternating minimization algorithm. Moreover, it can be seamlessly integrated with device velocity measurements obtained from onboard sensors (if available), for enhanced reliability. Here, AsyMov is thoroughly characterized by way of theory (Cramér-Rao bound), simulation, and experiments, implementing it on an IEEE 802.11ay testbed and testing it on multiple setups in the 60 GHz and 28 GHz bands, including moving human subjects. Numerical and experimental results show superior performance against state-of-the-art methods and are on par with scenarios featuring static ISAC devices.

eess.SP↗

Convolutional Spiking-based GRU Cell for Spatio-temporal Data

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.

cs.LG↗

Toward Multiband Sensing in FR3: Frequency Anisotropy Characterization and Non-Contiguous Bands Aggregation Algorithms

Frequency Range 3 (FR3) in the 7-24 GHz band will be the new spectrum for 6G wireless networks. The bandwidth availability and diversity of FR3 offer unprecedented opportunities for coherent multiband Integrated Sensing and Communications (ISAC), which aggregates the carrier phase information from multiple frequency bands to increase the sensing resolution to the cm-level. However, the frequency anisotropy of sensing targets over GHz-wide bands and the non-contiguity of the 6G spectrum, pose critical challenges to the application of existing multiband ISAC techniques. We present the first study on coherent multiband sensing in FR3. We experimentally characterize the frequency anisotropy of targets and propose new phase coherence metrics for multiband processing. Then, we analyze the impact of non-contiguous FR3 bands considered by 3GPP, and design a new algorithm to mitigate the resulting sensing artifacts, outperforming existing techniques. Our results represent a first step toward fully developing multiband ISAC for FR3.

eess.SP↗

Spatiotemporal Radar Gesture Recognition with Hybrid Spiking Neural Networks: Balancing Accuracy and Efficiency

Radar-based Human Activity Recognition (HAR) offers privacy and robustness over camera-based methods, yet remains computationally demanding for edge deployment. We present the first use of Spiking Neural Networks (SNNs) for radar-based HAR on aircraft marshalling signal classification. Our novel hybrid architecture combines convolutional modules for spatial feature extraction with Leaky Integrate-and-Fire (LIF) neurons for temporal processing, inherently capturing gesture dynamics. The model reduces trainable parameters by 88\% with under 1\% accuracy loss compared to baselines, and generalizes well to the Soli gesture dataset. Through systematic comparisons with Artificial Neural Networks, we demonstrate the trade-offs of spiking computation in terms of accuracy, latency, memory, and energy, establishing SNNs as an efficient and competitive solution for radar-based HAR.

cs.NE↗