Deep learning and convolutional neural networks allow achieving impressive performance in computer vision tasks, such as object detection semantic segmentation (SS). However, recent studies have shown evident weaknesses of models against adversarial perturbations. In a real-world scenario instead, like autonomous driving, more attention should be devoted to examples (RWAEs), which are physical objects (e.g., billboards printable patches) optimized the entire perception pipeline. This paper...

The existence of real-world adversarial examples (RWAEs) (commonly in the form patches) poses a serious threat for use deep learning models safety-critical computer vision tasks such as visual perception autonomous driving. This article presents an extensive evaluation robustness semantic segmentation (SS) when attacked with different types patches, including digital, simulated, and physical ones. A novel loss function is proposed to improve capabilities attackers inducing misclassification...

Computing platforms are evolving towards heterogeneous architectures including processors of different types and field programmable gate arrays (FPGAs), used as hardware accelerators for speeding up specific functions. The increasing capacity performance modern FPGAs, with their partial reconfiguration capabilities, have made them attractive in several application domains, space applications.This paper proposes a framework supporting the development safety-critical real-time systems that...

Next generation automotive applications require support for safe, predictable, and deterministic execution. The Logical Execution Time (LET) model has been introduced to improve the predictability correctness of time-critical applications. advent multicore architectures, together with need ensure time despite complex memory hierarchy hardware resources shared by cores, is an additional motivation use LET paradigm in conjunction a suitable scheduling access model. In this paper, we show how...

The study of parallel task models executed with predictable scheduling approaches is a fundamental problem for real-time multiprocessor systems. Nevertheless, to date, limited efforts have been spent in analyzing the combination partitioned and non-preemptive execution, which arguably one most schemes that can be envisaged handle tasks. This paper fills this gap by proposing an analysis sporadic DAG tasks under fixed-priority where computations corresponding nodes are non-preemptively...

Over the past few years, convolutional neural networks (CNNs) have proved to reach superhuman performance in visual recognition tasks. However, CNNs can easily be fooled by adversarial examples (AEs), i.e., maliciously crafted images that force predict an incorrect output while being extremely similar those for which a correct is predicted. Regular AEs are not robust input image transformations, then used detect whether AE presented network. Nevertheless, it still possible generate such...

The automotive industry is transitioning from federated, homogeneous, interconnected devices to integrated, heterogeneous, mixed-criticality systems (MCS). This leads challenges in achieving timing predictability techniques due access contention on shared resources, which can be mitigated using hardware-based spatial and temporal isolation techniques. Focusing the interconnect as point of for we propose AXI-REALM, a lightweight, modular, technology-independent, open-source real-time...

In this paper, we propose a method to perform empirical analysis of the loss landscape machine learning (ML) models. The is applied two ML models for scientific sensing, which necessitates quantization be deployed and are subject noise perturbations due experimental conditions. Our allows assessing robustness such effects as function precision under different regularization techniques -- crucial concerns that remained underexplored so far. By investigating interplay between performance,...

In the last decade, deep learning techniques reached human-level performance in several specific tasks as image recognition, object detection, and adaptive control. For this reason, is being seriously considered by industry to address difficult perceptual control problems safety-critical applications (e.g., autonomous driving, robotics, space missions). However, at moment, software poses a number of issues related safety, security, predictability, which prevent its usage systems. This letter...

When adopting multi-core systems for safety-critical applications, certification requirements mandate bounding the delays incurred in accessing shared resources. This is case of global memories, whose access often regulated by memory controllers optimized average-case performance and not designed to be predictable. As a consequence, worst-case bounds on result too pessimistic, drastically reducing advantage having multiple cores. paper proposes fine-grained analysis contention experienced...

FPGA-based system-on-chips (SoC) are powerful computing platforms to implement mixed-criticality systems that require both multiprocessing and hardware acceleration. Virtualization via hypervisor technologies is, de-facto, an effective technique allow the co-existence of multiple execution domains with different criticality levels in isolation upon same platform. Implementing such on SoC poses new challenges: one is accelerators deployed FPGA fabric belong but share common resources as a...

A stochastic modeling method is presented for the analysis of variability effects, induced by manufacturing process, on interconnect structures terminated general nonlinear loads. The technique based solution pertinent Telegrapher's equations in time domain means well-established Galerkin method, but now allows, first literature, inclusion loads with arbitrary I-V characteristics at terminals lines. transient obtained combining a finite-difference time-domain scheme. proposed validated and...

Engine control applications require the execution of tasks activated in relation to specific system variables, such as crankshaft rotation angle. To prevent possible overload conditions at high speeds, are designed vary their functionality (hence computational requirements) for different speed ranges. Modeling and analyzing a type poses new research challenges schedulability analysis that now being addressed real-time literature. This paper advances state art by presenting method computing...

Sharing resources in hierarchical real-time systems implemented with reservation servers requires the adoption of special budget management protocols that preserve bandwidth allocated to a specific component. In addition, blocking times must be accurately estimated guarantee both global feasibility all and local schedulability applications running on each This paper presents two new tests verify under fixed priority EDF schedulers. Reservation are BROE algorithm. A simple extension SRP...

This paper addresses the problem of providing spatial and temporal isolation between execution domains in a hypervisor running on an ARM multicore platform. Isolation is achieved by carefully managing two primary shared hardware resources today's platforms: last-level cache (LLC) DRAM memory controller. The XVISOR open-source Cortex A7 platform have been used as reference systems for purpose this work. Spatial partitioning LLC has implemented means coloring, which tightly integrated with...

AMBA AXI is a popular bus protocol that widely adopted as the medium to exchange data in field-programmable gate array system-on-chips (FPGA SoCs). The does not specify how conflicting transactions are arbitrated and hence design of arbiters left vendors adopt AXI. Typically, round-robin arbitration implemented ensure fair access by master nodes, for SoCs Xilinx. This paper addresses critical issue can arise when adopting under arbitration; specifically, presence with heterogeneous burst...

With heterogeneous multi-core platforms being crucial to execute the highly demanding workloads of modern applications, memory-access predictability remains a key issue for system's safety. Many solutions have been proposed over years, but none has applied on large scale. Nowadays, we are in front an unprecedented opportunity impact commercial platforms: Memory System Resource Partitioning and Monitoring (MPAM) specification by Arm, which describes different regulation mechanisms, presenting...