PMPaolo Montuschi

Paolo Montuschi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Paolo Montuschi has a total of 262 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    An OpenSSL Engine for Secure and Transparent Offloading of Cryptographic Operations to OP-TEE in Embedded IoT Platforms

    by Tina Sayarmoafi, Francesco Barchi, Lorenzo Bottaccioli, Andrea Acquaviva, Edoardo Patti, Paolo Montuschi, Luca Barbierato

    Embedded IoT platforms are frequently deployed in hostile or physically exposed environments, where compromise of the operating system is a realistic threat. In conventional deployments, OpenSSL executes entirely in user space, leaving cryptographic keys and intermediate material potentially exposed in the presence of a compromised OS or privileged malware. This work presents a portable OpenSSL engine that enables secure and transparent offloading of cryptographic operations to OP-TEE, a software stack leveraging a Trusted Execution Environment (TEE) to confine cryptographic key material and security-critical computations within secure-world memory without modifying existing applications or altering the OpenSSL EVP interface. The proposed engine uses a GlobalPlatform Client API implementation as an underlying communication layer and supports both Copy-Based (CB) data transfer and a Shared-Memory (SM) mode, enabling performance tuning under the resource constraints typical of embedded IoT platforms. An experimental evaluation on an NXP i.MX7 industrial gateway shows that secure-world execution introduces bounded overhead dominated by REE–TEE transitions and memory-management operations. For bandwidth-intensive primitives such as SHA-256, SM mode improves throughput by approximately 10–15% over CB transfers. For symmetric encryption algorithms such as AES-256-CBC, SM communication provides a substantial throughput improvement of approximately 40%, indicating that data-movement overhead plays a significant role on embedded platforms. In contrast, asymmetric primitives (e.g. RSA-1024) remain largely insensitive to transfer optimisations because they operate on small, fixed-size operands, making the overall execution time dominated by invocation overheads and internal big-number computations rather than data movement.

  • Runtime Feature Compression for Adaptive Keyword Spotting on Embedded Systems

    by Valentino Peluso, Andrea Calimera, Enrico Macii, Paolo Montuschi

    Voice user interfaces rely on keyword spotting (KWS) to detect wake-word commands, enabling low-power devices to switch from drowsy to active states and initiate more complex tasks. In embedded systems, KWS combines handcrafted acoustic features extraction with lightweight neural network (NN) classifiers to achieve accurate detection within strict resource constraints. Adapting KWS to time-varying energy budgets requires optimization strategies that operate at runtime. Most existing approaches adjust the complexity of the neural model but overlook that a substantial amount of latency, and thus energy consumption, is due to feature extraction, which remains unaffected by model scaling. This work introduces runtime feature compression (RFC), a dynamic rescaling strategy that modulates the workload of the entire KWS pipeline. RFC promotes the hop-length parameter of the short-time Fourier transform (STFT) as a runtime control knob to adjust the number of time frames in speech features, allowing a single model to operate across multiple latency modes. To support this flexibility, we introduce two training-time techniques: HopAugment, a data augmentation scheme that exposes the model to variable hop lengths during training, and masked layers, which preserve consistent activation statistics during training and inference under compressed feature settings. Evaluations on four KWS datasets using the TC-ResNet model family show that RFC outperforms model scaling techniques, offering a wider range of latency-accuracy tradeoffs. RFC achieves up to 31.8% lower latency without accuracy degradation, or up to 0.30% higher accuracy within equivalent latency bounds. That proves RFC improves adaptability in energy-constrained IoT speech interfaces. A set of ablation studies further demonstrates the robustness of RFC by evaluating the role of its training components, batching strategies, ability to preserve accuracy with a shared weight set, scalability across operating modes, and applicability to different model architectures.

  • Open Access

    The inNuCE Research Infrastructure and the Neuromorphic MLOps for AIoT Prototyping

    by Gianvito Urgese, Vittorio Fra, Andrea Pignata, Giuseppe Fanuli, Walter Gallego Gomez, Riccardo Pignari, Michelangelo Barocci, Benedetto Leto, Salvatore Tilocca, Nicola Cassetta, et al.

    Neuromorphic computing promises significant improvements in latency and energy efficiency for machine intelligence at the edge. However, its adoption in the IoT domain is still limited by the heterogeneity of the hardware (HW), the immaturity of the toolchains, and the poor reproducibility of experiments. This article sets out the inNuCE research infrastructure (inNuCE RI), a two-pillar facility composed of a physical laboratory (inNuCE Lab) and a cloud-based heterogeneous prototyping platform (inNuCE HPP). The purpose of the inNuCE RI is to enable developers to prototype, evaluate, and compare neuromorphic and conventional end-to-end digital solutions. From a methodological perspective, we formalize the adaptation of machine learning operations (MLOps) to event-driven sensing and brain-inspired computation as neuromorphic MLOps (NMLOps). We illustrate how inNuCE RI instantiates NMLOps through containerized toolchains orchestrated with Kubernetes and Slurm-managed heterogeneous resources (neuromorphic chips, FPGAs, GPUs, and MCUs). The approach is analyzed on representative Artificial Intelligence of Things (AIoT) use cases, including human activity recognition (HAR), Braille reading, event-based gesture recognition, Hi-Co semantization of memories, navigation tracking, and constraint satisfaction problems. The development of inNuCE RI has been driven by the need to facilitate the transition from prototype (in nuce) to engineered AIoT systems for lower entry barriers and enforce reproducibility. This paves the way for future system-of-systems engineering.


  • Open Access

    Multimodal Fusion of Face and Gait for Person Identification in Automotive Applications

    by Federico Boscolo, Fabrizio Lamberti, Paolo Montuschi, Mario Testa

    Smart and secure access to vehicles is a crucial aspect of the evolving automotive industry. This paper focuses on the development of an end-to-end multimodal biometric recognition framework that identifies people walking towards a vehicle from an RGB video feed. The framework is based on a deep-learning pipeline for person detection and tracking, face and gait feature extraction, and fusion of the two modalities at the score and feature level. Traditional face recognition systems can suffer from variations in lighting and occlusions. In order to deal with these issues, the proposed framework integrates face and gait features with the aim to enhance accuracy. The pipeline is modular, enabling seamless integration of new models for each step of person identification without the need for additional training. Baseline face and gait recognition models, as well as score- and feature-level fusion techniques are evaluated on subsets of the CASIA-A and CASIA-B datasets. Experimental results show that weighted mean score-level fusion significantly improves both Rank-1 accuracy and verification accuracy (TAR@FAR=10−5) over unimodal baselines. Overall, reported work provides insights into current limitations and suggests directions for future research about secure identity verification in vehicles.


  • Dealing With Challenged IoT Networks in Hierarchical Federated Learning

    by Alessio Sacco, Doriana Monaco, Guido Marchetto, Paolo Montuschi

    Federated learning (FL) has revolutionized the way in which mobile devices and Internet of Things (IoT) can share common knowledge in data analytics. However, some challenges arise when dealing with heterogeneous and challenged networks, especially in gradient synchronization. For example, some clients (referred to as stragglers) may take much longer to report their output than other nodes. Current solutions addressing the straggling problems either propose a distributed coordination (but introduce new synchronization issues) or deadline-based approaches to discard clients after a fixed deadline (but introduce the problem of determining a suitable deadline). To this end, we propose to set a dynamic deadline in which the central server selects the best IoT nodes via an online learning approach based on predicting the response time of each client. Moreover, to further mitigate synchronization and scalability issues, we also consider a hierarchical approach in which clients send model parameters to intermediate aggregation edge servers. Our results demonstrate that this approach can lower network overhead by 78% compared to the widely adopted FedAvg and 49% to the best alternative. At the same time, the model accuracy is preserved, and the training time in challenged networks is reduced by 52% w.r.t. FedAvg and 32% w.r.t. recent solutions.


  • A Configurable Floating-Point Fused Multiply-Add Design With Mixed Precision for AI Accelerators

    by Farzad Niknia, Ziheng Wang, Shanshan Liu, Pedro Reviriego, Zhen Gao, Paolo Montuschi, Fabrizio Lombardi

    Hardware accelerators for deep learning in artificial intelligence applications must often meet stringent constraints for accuracy and throughput. In addition to architecture/algorithm improvements, high performance computational techniques such as mixed precision are also required. In this paper, a floating-point (FP) fused multiply-add (FMA) unit supporting mixed/multiple precision is proposed. A wide range of conventional FP formats (such as half and single) as well as emerging formats (including E4M3, E5M2, DLFloat, BFLoat16 and TF32) are supported in the proposed design. In addition to all these formats, the proposed design is flexible in manipulating the exponent and mantissa lengths for 8, 16 and 32-bit FP numbers based on the needs of an application. The proposed FMA can be configured to support either multiple normal FMA operations, or alternatively mixed precision in ASIC. It is fully pipelined and in each cycle, the input bit streams are processed based on the provided configuration, so independent of the previous cycles. For normal FMA operations, the proposed design utilizes sharing of resources to parallelize multiple operations based on the available hardware and required precision. For mixed precision the FMA accumulates the lower precision dot products into higher precision to avoid overflow/underflow. It improves computational accuracy by adding all possible dot products at the same time while decreasing the number of rounding operations to prevent rounding errors. An innovative method to accumulate the dot products and the aligned addend is also proposed. By considering tradeoffs between reusing the available hardware and removing unnecessary complex units, a more efficient and flexible design is attained in terms of hardware metrics and supported different precision computation compared to other designs found in the technical literature. Extensive simulation results for comparative analysis are provided.


  • In-Memory Computing: The Emerging Computing Topic in the Post-von Neumann Era

    by Paolo Montuschi, Yuan-Hao Chang, Vincenzo Piuri

    In-memory computing is a paradigm to break the increasing gap between the processor and memory speeds by performing computation inside or near the memory. Here, we aim to stimulate the curiosity of readers toward this new, emerging area.


  • Automatic Layer Freezing for Communication Efficiency in Cross-Device Federated Learning

    by Erich Malan, Valentino Peluso, Andrea Calimera, Enrico Macii, Paolo Montuschi

    Federated Learning (FL) is a collaborative machine learning paradigm where network-edge clients train a global model under the orchestration of a central server. Unlike traditional distributed learning, each participating client keeps its data locally, ensuring privacy protection by default. However, state-of-the-art FL implementations suffer from massive information exchange between clients and the server. This issue prevents the adoption in constrained environments, typical of the Internet-of-Things domain, where the communication bandwidth and the energy budget are severely limited. To achieve higher efficiency at scale, the future of FL calls for additional optimizations to reach high-quality learning capability with lower communication pressure. To address this challenge, we propose Automatic Layer Freezing (ALF), an embedded mechanism that gradually drops a growing portion of the model out of the training and synchronization phases of the learning loop, reducing the volume of exchanged data with the central server. ALF monitors the evolution of model updates and identifies layers that have reached a stable representation, where further weight updates would have minimal impact on accuracy. By freezing these layers, ALF achieves substantial savings in communication bandwidth and energy consumption. The proposed implementation of the ALF mechanism is compatible with any FL strategy, requiring minimal effort and without interfering with existing optimizations. The extensive experiments conducted using a representative set of FL strategies applied to two image classification tasks show that ALF improves the communication efficiency of the baseline FL implementations, ensuring up to 83.91% of data volume savings with no or marginal losses of accuracy.


  • Open Access

    Proceedings of the IEEE Publication Information

    by D.J. Allstot, John Anderson, John Baillieul, Gert Cauwenberghs, Raja Chatila, Michael Fang, Gerhard Fettweis, Cynthia Furse, Georges Gielen, Lawrence J. Hall, et al.

  • Dynamic Decision Tree Ensembles for Energy-Efficient Inference on IoT Edge Nodes

    by Francesco Daghero, Alessio Burrello, Enrico Macii, Paolo Montuschi, Massimo Poncino, Daniele Jahier Pagliari

    With the increasing popularity of Internet of Things (IoT) devices, there is a growing need for energy-efficient Machine Learning (ML) models that can run on constrained edge nodes. Decision tree ensembles, such as Random Forests (RFs) and Gradient Boosting (GBTs), are particularly suited for this task, given their relatively low complexity compared to other alternatives. However, their inference time and energy costs are still significant for edge hardware. Given that said costs grow linearly with the ensemble size, this paper proposes the use of dynamic ensembles, that adjust the number of executed trees based both on a latency/energy target and on the complexity of the processed input, to trade-off computational cost and accuracy. We focus on deploying these algorithms on multi-core low-power IoT devices, designing a tool that automatically converts a Python ensemble into optimized C code, and exploring several optimizations that account for the available parallelism and memory hierarchy. We extensively benchmark both static and dynamic RFs and GBTs on three state-of-the-art IoT-relevant datasets, using an 8-core ultra-lowpower System-on-Chip (SoC), GAP8, as the target platform. Thanks to the proposed early-stopping mechanisms, we achieve an energy reduction of up to 37.9% with respect to static GBTs (8.82 uJ vs 14.20 uJ per inference) and 41.7% with respect to static RFs (2.86 uJ vs 4.90 uJ per inference), without losing accuracy compared to the static model.