Search

Simulink is a widely used model-driven design environment for supporting the simulation and code generation of embedded applications. To improve the quality of the code generated from Simulink models, state-of-the-art code generators employ various high-level optimizations, like eliminating local variables. However, they overlook the compatibility between code and the low-level processor architecture, especially the instruction pipeline. Consequently, instruction pipeline stalls occur frequently, leading to additional delays in instruction execution, as well as limited efficiency for deployed the embedded software. In this article, we propose Mercury, an instruction pipeline aware code generator for Simulink models which utilizes data dependencies between actors to decrease the instruction pipeline stalls of the generated code. First, Mercury collects data dependencies through model dataflow traversal and records the property of each actor. Then, Mercury approximately estimates the execution latency of required instructions fetched from corresponding actors and uses a topology-based method to obtain candidate actors for code synthesis. Finally, Mercury adopts the least penalty priority to iteratively select the most suitable actor for code synthesis and releases data dependencies with its subsequent actors. We implemented and evaluated Mercury on benchmark Simulink models (Su et al., 2021) as well as a real industrial model. Compared to the official tool Simulink Embedded Coder and the state-of-the-art academic tool DFSynth, Mercury outperformed them by 9.7%–33.4% and 9.2%–59.4% in terms of the execution time of the generated code across different architectures, respectively. The statistics also demonstrate that the generated code of Mercury increases utilization of pipeline slots by 11.0%–37.1% and 10.6%–50.0%, respectively. [ABSTRACT FROM AUTHOR]

In recent years, since edge computing has become more and more popular, its security issues have become apparent and have received unprecedented attention. Thus, the current research concentrates on security not only regarding devices such as PCs, smartphones, tablets, and IoTs, but also the automobile industry. However, since attack vectors have become more sophisticated than ever, we cannot just protect the zone above the system software layer in a certain operating system, such as Linux, for example. In addition, the challenges in IoT devices, such as power consumption, performance efficiency, and authentication management, still need to be solved. Since most IoT devices are controlled remotely, the security regarding system maintenance and upgrades has become a big issue. Therefore, a mechanism that can maintain IoT devices within a trusted environment based on localhost or over-the-air (OTA) will be a viable solution. We propose a mechanism called STBEAT, integrating an open-source project with ARM TrustZone to solve the challenges of upgrading the IoT system and updating system files more safely. This paper focuses on the ARMv7 architecture and utilizes the security stack from TrustZone to OP-TEE under the STM32 board package, and finally obtains the security key from the trusted application, which is used to conduct the cryptographic operations and then install the newer image on the MMC interface. To sum up, we propose a novel software update strategy and integrated ARM TrustZone security extension to beef up the embedded ecosystem. [ABSTRACT FROM AUTHOR]

LiDAR sensors are extensively used in autonomous vehicles and their optimal use is a critical concern. In this paper, we propose an embedded software architecture for multiple LiDAR sensors, i.e., a fusion system that acts as an embedded system for processing data from multiple LiDAR sensors. The fusion system software comprises multiple clients and a single server. The client and server are connected through inter-process communication. Multiple clients create processes to process the data from each LiDAR sensor via a multiprocessing method. Our approach involves a scheduling method for efficient multiprocessing. The server uses multithreading to optimize the internal functions. For internal communication within the fusion system, multiple clients and a single server are connected using the socket method. In sequential processing, the response time increases in proportion to the number of connected LiDAR sensors. By contrast, in the proposed software architecture, the response time decreases in inverse proportion to the number of LiDAR sensors. As LiDAR sensors become increasingly popular in the field of autonomous driving, the results of this study can be expected to make a substantial contribution to technology development in this domain. [ABSTRACT FROM AUTHOR]

This paper describes a study on the power and energy consumption estimation models that have been defined to facilitate the development of ultra-low power embedded applications. During the study, various measurements have been carried out on the instruction and application level to challenge the models against empirical data. The study has been performed on the multicore heterogeneous hardware platform developed for ultra-low power Digital Signal Processors (DSP) applications. The final goal was to develop a tool that can provide insight into power dissipation during the execution of embedded applications, so that one can refactor the source code in an energy-efficient manner, or ideally to develop an energy-aware C compiler. The side effect of the research presents interesting insight into how the custom hardware architecture influences power dissipation. The selected platform has been chosen simply because it represents R&D state of the art ultra-low power hardware used in hearing aids. The presented solution has been developed and tested in an Eclipse environment using Java programming language. [ABSTRACT FROM AUTHOR]

ObjectiveConventional brain-computer interfaces (BCIs) are often expensive, complex to operate, and lack portability, which confines their use to laboratory settings. Portable, inexpensive BCIs can mitigate these problems, but it remains unclear whether their low-cost design compromises their performance. Therefore, we developed a portable, low-cost BCI and compared its performance to that of a conventional BCI.MethodsThe BCI was assembled by integrating a custom electroencephalogram (EEG) amplifier with an open-source microcontroller and a touchscreen. The function of the amplifier was first validated against a commercial bioamplifier, followed by a head-to-head comparison between the custom BCI (using four EEG channels) and a conventional 32-channel BCI. Specifically, five able-bodied subjects were cued to alternate between hand opening/closing and remaining motionless while the BCI decoded their movement state in real time and provided visual feedback through a light emitting diode. Subjects repeated the above task for a total of 10 trials, and were unaware of which system was being used. The performance in each trial was defined as the temporal correlation between the cues and the decoded states.ResultsThe EEG data simultaneously acquired with the custom and commercial amplifiers were visually similar and highly correlated ( ρ = 0.79). The decoding performances of the custom and conventional BCIs averaged across trials and subjects were 0.70 ± 0.12 and 0.68 ± 0.10, respectively, and were not significantly different.ConclusionThe performance of our portable, low-cost BCI is comparable to that of the conventional BCIs.SignificancePlatforms, such as the one developed here, are suitable for BCI applications outside of a laboratory.

Nowadays, deep learning has become one of the most popular solutions for computer vision, and it has also included the Edge. It has influenced the System-on-Chip (SoC) vendors to integrate accelerators for inference tasks into their SoCs, including NVIDIA, NXP, and Texas Instruments embedded systems. This work explores the performance of the NXP i.MX8M Plus Neural Processing Unit (NPU) as one of the solutions for inference tasks. For measuring the performance, we propose an experiment that uses a GStreamer pipeline for inferring license plates, which is composed of two stages: license plate detection and character inference. The benchmark takes execution time and CPU usage samples within the metrics when running the inference serially and parallel. The results show that the key benefit of using the NPU is the CPU freeing for other tasks. After offloading the license plate detection to NPU, we lowered the overall CPU consumption by 10x. The performance obtained has an inference rate of 1 Hz, limited by the character inference.

Face recognition payment is a new type of payment method, with AI face recognition technology as the core, and its speed and convenience are more in line with the users' payment habits. However, the face is a biological feature with weak privacy, and the protection of user information security is particularly important. At present, face payment technology still has security risks, and the data transmitted during the transaction process are vulnerable to attacks. Aiming at the security problems in the payment process, a payment system that is jointly encrypted by the SM4 algorithm and the face liveness detection algorithm was proposed in this paper, which supports a variety of communication methods. The hardware platform adopts an octa-core 64-bit ARM processor with a main frequency of 1.8 GHz, which has powerful computing and processing capabilities. Based on the Android intelligent operating system, the development environment is more secure and convenient. It is also equipped with a liveness detection 3D structured light camera, which dynamically collects face information and accurately analyzes the characteristics of living bodies. Through the data encryption and decryption test and face performance index detection, the expected effect of the system was achieved, which greatly improved the performance of the face payment system currently studied. The SM4 encryption algorithm improved the running rate of encrypted data and the security of face transaction data transmission, the face detection algorithm improved the accuracy of living body feature recognition, and the payment system effectively improved the accuracy and security of face payment. [ABSTRACT FROM AUTHOR]

Trends in the automotive industry confirm that the demand for testing of embedded systems, especially advanced driver assistance systems (ADAS), will grow dramatically in the near future. This paper proposes a new solution that automates the detection of software defects in embedded systems. The solution consists of a data-driven sampling algorithm to intelligently sample the testing space by sequentially generating test cases. Moreover, it segregates different defects from each other and identifies the signals that trigger each. The results are compared against other automated methods for defect identification and analysis, and it is found that this novel solution is able to identify defects more rapidly. In addition, it correctly separates defects and reliably reproduces each distinct defect. [ABSTRACT FROM AUTHOR]

Devices that populate the Internet of Things (IoT) are typically constrained with respect to energy consumption. When the data that are processed, stored and/or communicated by these devices need to be secured, low-energy security mechanisms have to be designed and implemented. Related work mainly concentrates either on low-energy security algorithms and protocols, or on low-energy wireless communication. However, it is important for system developers to take into account the overall energy consumption of the IoT system when making design choices. Therefore, this work presents an in-depth analysis of the energy consumption of IoT devices that provide end-to-end secure communication and digital signatures. The paper follows a granular approach, profiling and measuring each individual contribution to the overall energy consumption, including the computation of cryptographic operations as well as the wireless transmission of messages in cryptographic protocols. The paper also calculates the minimal time period of a secure communication session in order to minimize the energy impact of the session's setup phase and thus minimize the overall average power consumption. The goal of this work is to provide assistance in the selection of a suitable wireless communication standard and cryptographic cipher suite for building end-to-end secure IoT applications. [ABSTRACT FROM AUTHOR]

Many embedded systems operate under real-time conditions. For them, the worst case execution time (WCET) is a crucial information to check whether the software implementation meets the requirements. In modern microcontrollers, this WCET can often no longer be evaluated statically since these processors include too many unpredictable components (cache, bus arbitration, $\ldots$ ,). Here, hybrid WCET estimation comes into play, where measured traces are used to empirically compute bounds for the WCET. For meaningful estimations, it is required to put each measured segment into a corresponding execution context. In this contribution, we present a method that allows to gather such context-sensitive statistics online in hardware. This enables arbitrary long measurement intervals with a precision that compares to state-of-the-art offline tools. [ABSTRACT FROM AUTHOR]

The success of CubeSat space missions depends on the ability to perform properly in a harsh environment. A key component in space missions is the flight software, which manages all of the processes executed by the satellite on its onboard computer. Literature shows that CubeSat missions suffer high infant mortality, and many spacecraft failures are related to flight software errors, some of them resulting in complete mission loss. Extensive operation testing is the primary technique used by CubeSats developers to ensure flight software quality and avoid such failures. The “New Space” requirements pressure to add “agility” to the software development, which could limit the capacity to test. While advanced and beneficial software testing techniques are found in the software engineering field, CubeSat software solutions mostly rely on unit testing, software in the loop simulation, and hardware in the loop simulation. In this work, fuzz testing techniques were developed, implemented, and evaluated as a manner to expedite operational testing of CubeSats while maintaining their completeness. The impact of the tools was evaluated by using the three new 3U CubeSats under development at the University of Chile. We identified twelve bugs not covered by classic testing strategies in less than three days. These failures were reported, fixed, and characterized by the developers in eight sprint sessions. Our results indicate that fuzz testing improved the completeness of flight software testing through automation and with almost no development interruption. Although our approach has been tested on the SUCHAI flight software, it applies to systems that follow a similar architecture.

Assisted bilateral rehabilitation has been proven to help patients improve their paretic limb ability and promote motor recovery, especially in upper limbs, after suffering a cerebrovascular accident (ACV). Robotic-assisted bilateral rehabilitation based on sEMG-driven control has been previously addressed in other studies to improve hand mobility; however, low-cost embedded solutions for the real-time bio-cooperative control of robotic rehabilitation platforms are lacking. This paper presents the RobHand (Robot for Hand Rehabilitation) system, which is an exoskeleton that supports EMG-driven assisted bilateral by using a custom-made low-cost EMG real-time embedded solution. A threshold non-pattern recognition EMG-driven control for RobHand has been developed, and it detects hand gestures of the healthy hand and replicates the gesture on the exoskeleton placed on the paretic hand. A preliminary study with ten healthy subjects is conducted to evaluate the performance in reliability, tracking accuracy and response time of the proposed EMG-driven control strategy using the EMG real-time embedded solution, and the findings could be extrapolated to stroke patients. A systematic review has been carried out to compare the results of the study, which present a 97% of overall accuracy for the detection of hand gestures and indicate the adequate time responsiveness of the system.

QR code payment plays an indispensable role in the mobile payment market, and the security of scanning codes has always been a problem in the field of information security. Static QR codes are easily copied and replaced, and there are huge security loopholes. The QR code payment in a closed system still faces security challenges. In order to solve the security problem of QR code payment, we have studied dynamic QR code payment system that supports SM2, SM3, and SM4 cryptographic algorithms, which can realize QR code scanning and scanned transactions, UnionPay cloud QuickPass transactions, etc., and generate dynamic QR code information in real time during the transaction process, one order and one code. Through dynamic algorithm distribution, the randomness and uniqueness of QR code generation are guaranteed, and it is suitable for multi-scene application transactions. The algorithm correctness test result shows that the system has achieved the expected effect. The performance test results show that the hardware of the security module implements the algorithm flow and improves the payment performance. Compared with some other algorithms, the processing time is shorter, the running speed is faster, and the system is more secure.

Object detection is an essential technology for surveillance systems, particularly in areas with a high risk of accidents such as railway level crossings. To prevent future collisions, the system must detect and track any object that passes through the monitored area with high accuracy, and this process must be performed fulfilling real-time specifications. In this work, an edge IoT HW platform implementation capable of detecting and tracking objects in a railway level crossing scenario is proposed. The response of the system has to be calculated and sent from the proposed IoT platform to the train, so as to trigger a warning action to avoid a possible collision. The system uses a low-resolution 3D 16-channel LIDAR as a sensor that provides an accurate point cloud map with a large amount of data. The element used to process the information is a custom embedded edge platform with low computing resources and low-power consumption. This processing element is located as close as possible to the sensor, where data is generated to improve latency, privacy, and avoid bandwidth limitations, compared to performing processing in the cloud. Additionally, lightweight object detection and tracking algorithm is proposed in this work to process a large amount of information provided by the LIDAR, allowing to reach real-time specifications. The proposed method is validated quantitatively by carrying out implementation on a car road, emulating a railway level crossing.

With the rise of IoT and edge computing, deploying neural networks (NNs) on low-power edge computing devices is drawing more and more attention. In NNs, convolutional layers take up the majority of the computing cycles, especially when NNs are implemented on ARM processors. Therefore, it is necessary to optimize the convolutional implementation on ARM Cortex-M MCUs. This paper proposes an efficient im2row-based fast convolution algorithm with two innovations. First, a novel im2row method for reusing the data of adjacent convolutional windows is presented. This method utilizes a reusable im2row buffer for data reuse, significantly reducing the amount of data copied during im2row and improving efficiency. Second, in algorithm implementation, a q7_t to q15_t data type extension technique that avoids data reordering is employed. This technique eliminates data reordering instructions, thus reducing the runtime of the algorithm. We evaluate our algorithm in separate convolutional layers and NNs. The results for convolutional layers show that, compared to baseline, the proposed algorithm speeds up the convolutional layer by an average of $1.42\times $ , and the maximum speedup is up to $2.9\times $ . Experiments on different NNs demonstrate that our algorithm can speed up the overall NN by up to $2.15\times $ .

In the past decade, Deep Neural Networks (DNNs), e.g., Convolutional Neural Networks, achieved human-level performance in vision tasks such as object classification and detection. However, DNNs are known to be computationally expensive and thus hard to be deployed in real-time and edge applications. Many previous works have focused on DNN model compression to obtain smaller parameter sizes and consequently, less computational cost. Such methods, however, often introduce noticeable accuracy degradation. In this work, we optimize a state-of-the-art DNN-based video detection framework—Deep Feature Flow (DFF) from the cloud end using three proposed ideas. First, we propose Asynchronous DFF (ADFF) to asynchronously execute the neural networks. Second, we propose a Video-based Dynamic Scheduling (VDS) method that decides the detection frequency based on the magnitude of movement between video frames. Last,we propose Spatial Sparsity Inference, which only performs the inference on part of the video frame and thus reduces the computation cost. According to our experimental results, ADFF can reduce the bottleneck latency from 89 to 19 ms. VDS increases the detection accuracy by 0.6% mAP without increasing computation cost. And SSI further saves 0.2 ms with a 0.6% mAP degradation of detection accuracy. [ABSTRACT FROM AUTHOR]

Maritime Autonomous Surface Ships (MASS), an emerging area of digital advancement in shipping and shipbuilding industries, presents a different legal paradigm from that of existing ships. Existing maritime-related industries, including shipping, shipbuilding, and logistics, based on large hardware called ships, are rapidly changing into highly autonomous software-centered structures. This study is focused on analyzing the legal issues in preparation for MASS' commercial operations in the future by applying comparative methods centered on the Republic of Korea and the United Kingdom. The study's results contribute to the criteria for the manufacturing responsibility of autonomous ship-embedded software and a desirable legal policy improvement plan. Various legal issues and macro legal policy directions related to software product liability were identified. The study also presents concrete implementation strategies to achieve an ideological harmony and a balance between equitable damage relief and the advancement of related technologies to ensure maritime safety in the maritime industry. Based on the issues identified and their legal policy alternatives, it is hoped that the institutional ideal of product liability, which promotes technological advancement and protects consumer rights, is realized in the software domain as well. [ABSTRACT FROM AUTHOR]

Real-time globally consistent GPS tracking is critical for an accurate localization and is crucial for applications such as autonomous navigation or multi-robot mapping. However, under challenging environment conditions such as indoor/outdoor transitions, GPS signals are partially available or not consistent over time. In this paper, a real-time tracking system for continuously locating emergency response agents in challenging conditions is presented. A cooperative localization method based on Laser-Visual-Inertial (LVI) and GPS sensors is achieved by communicating optimization events between a LiDAR-Inertial-SLAM (LI-SLAM) and Visual-Inertial-SLAM (VI-SLAM) that operate simultaneously. The estimation of the pose assisted by multiple SLAM approaches provides the GPS localization of the agent when a stand-alone GPS fails. The system has been tested under the terms of the MALIN Challenge, which aims to globally localize agents across outdoor and indoor environments under challenging conditions (such as smoked rooms, stairs, indoor/outdoor transitions, repetitive patterns, extreme lighting changes) where it is well known that a stand-alone SLAM will not be enough to maintaining the localization. The system achieved Absolute Trajectory Error of 0.48%, with a pose update rate between 15 and 20 Hz. Furthermore, the system is able to build a global consistent 3D LiDAR Map that is post-processed to create a 3D reconstruction at different level of details. [ABSTRACT FROM AUTHOR]

Much of the previous research has been done on distributed software transactional memories targeting data centers in Internet clouds, which resulted in nondeterministic and nonrealtime middleware solutions mainly written in Java and C++. On the other hand, embedded systems based on the Internet of Things at the edge of the Internet, such as smart homes, cars, etc., need to operate in realtime, and should, therefore, be deterministic. In order to be smart, these systems use machine learning, and nowadays Python is becoming a leading language in this venue, too. This is the first paper that presents a distributed software transactional memory that is at the same time: (i) fault tolerant, (ii) deterministic, (iii) based on Python, and (iv) extended from a formally verified root. The presented solution consists of a pair of master-slave transaction coordinators and a set of replicated data servers and targets small-to-medium edge networks. Besides intelligent embedded systems, based on the Internet of Things, it can be also used in a wide range of application domains, from SCADA systems to large-scale simulations. The experimental results, presented in the paper, show a superlinear growth of the system throughput as the workload changes from write-only to read-only.

Abstract A keyword spotting algorithm implemented on an embedded system using a depthwise separable convolutional neural network classifier is reported. The proposed system was derived from a high-complexity system with the goal to reduce complexity and to increase efficiency. In order to meet the requirements set by hardware resource constraints, a limited hyper-parameter grid search was performed, which showed that network complexity could be drastically reduced with little effect on classification accuracy. It was furthermore found that quantization of pre-trained networks using mixed and dynamic fixed point principles could reduce the memory footprint and computational requirements without lowering classification accuracy. Data augmentation techniques were used to increase network robustness in unseen acoustic conditions by mixing training data with realistic noise recordings. Finally, the system’s ability to detect keywords in a continuous audio stream was successfully demonstrated.

The aim of this paper and research was to analyse the efficiency of the compiler-generated code for the graphics library and to present results obtained by optimization for the MIPS (Million Instructions Per Second) architecture. Libpng is the official Portable Network Graphics reference library for use in applications that read, create, and manipulate PNG (Portable Network Graphics) raster image files. Given the data structure in the PNG files, as well as the capabilities of the MIPS instruction set, it was expected that significant improvements could be made. Graphic library libpng is optimized by using MIPS instruction set extension and tested on MIPS Malta 74K platform. Test results show, that by using MIPS optimization test, execution times are substantially improved. Our libpng optimization have achieved performance increase of 10 %–78 % depending on optimized routine.

Digital active disturbance rejection controllers (DADRC) have a wide scope as they have inherent advantages such as: easy hardware implementation, the entire system's mathematical model it is not necessary, is a linear controller dealing with uncertainty and is intended to ensure that energy savings are achieved. This document provides a procedure for designing a DADRC for a dc motor drive from the perspective of a Generalized Proportional Integral observer and differential flatness. The angular speed of a dc motor drive is the main purpose of the controller. The proposed DADRC was implemented using floating point operations on a field programmable gate array Artix-7™ using a Nexys 4 board. The system is subject to current demands, parameter variations, dc power supply changes and external load torque variations. The coupling of buck converter and dc motor resulting in external and internal disturbances that are actively counteracts the linear controller, as shown in the results.

Transiently-powered devices rely solely on energy harvesters. Such devices typically use capacitors to store the harvested energy, but recent systems employ supercapacitors to store energy for extended operations. In a supercapacitor-based transiently-powered system, the energy-efficient estimation of the state-of-charge (SoC) of a supercapacitor is critical for practical use of the system, due to its tight energy budget. Conventional voltage-based schemes for capacitors do not provide accuracy in SoC estimation for supercapacitors. Also, the supercapacitor-specific SoC estimation which exploits its charge redistribution characteristics provides an accuracy, but the scheme demands significant overhead, and thus, is not applicable to transiently-powered systems, such as wireless sensor nodes. In this article, we specify three requirements for estimating supercapacitor SoC that should be met to function in transiently-powered systems. We then propose a scheme that meets those requirements. The proposed scheme does not require additional hardware, has a low computation cost, and operates when the system is intermittently powered. Thus, the scheme fits the energy-efficient operation of transiently-powered systems. We implemented the proposed scheme in real hardware and evaluated its functionality and accuracy. The proposed scheme estimated the SoC of supercapacitors with high accuracy, in various configurations and use scenarios, while guaranteeing operations in a typical energy harvesting environment. [ABSTRACT FROM AUTHOR]

The increasing complexity of embedded systems renders software verification more complex, requiring monitoring and formal techniques, like model-checking. However, to use such techniques, system engineers usually need formal expertise to express the software requirements in a formal language. To facilitate the use of model-checking tools by system engineers, our approach uses a UML model interpreter through which the software requirements can directly be expressed in UML as well. Formal requirements are encoded as UML state machines with the transition guards written in a specific observation language, which expresses predicates on the execution of the system model. Each such executable UML specification can model either a Büchi automaton or an observer automaton, and is synchronously composed with the system, to follow its execution during model-checking. Formal verification can continue at runtime for all deterministic observer automata used during offline verification by deploying them on real embedded systems. Our approach has been evaluated on multiple case studies and is illustrated, in this paper, through the user interface model of a cruise-control system. The automata-based verification results are in line with the verification of the equivalent LTL properties. The runtime overhead during monitoring is proportional to the number of monitors. [ABSTRACT FROM AUTHOR]

We report on the development, installation, and operation of the first three of seven stations deployed at the ARIANNA site's pilot Hexagonal Radio Array (HRA) in Antarctica. The primary goal of the ARIANNA project is to observe ultrahigh energy (> 100PeV) cosmogenic neutrino signatures using a large array of autonomous stations, each 1 km apart on the surface of the Ross Ice Shelf. Sensing radio emissions of 100 MHz to 1 GHz, each station in the array contains RF antennas, amplifiers, 1.92 G-sample/s, 850 MHz bandwidth signal acquisition circuitry, pattern-matching trigger capabilities, an embedded CPU, 32 GB of solid-state data storage, and long-distance wireless and satellite communications. Power is provided by the sun and buffered in LiFePO4 storage batteries, and each station consumes an average of 7 W of power. Operation on solar power has resulted in ≥58% per calendar-year live-time. The station's pattern-trigger capabilities reduce the trigger rates to a few milli-Hertz with 4-sigma voltage thresholds while retaining good stability and high efficiency for neutrino signals. The timing resolution of the station has been found to be 0.049 ns, RMS, and the angular precision of event reconstructions of signals bounced off of the sea-ice interface of the Ross Ice Shelf ranged from 0.14 to 0.17°.

Deep Neural Networks (DNNs) have gained widespread popularity in different domain applications due to their dominant performance. Despite the prevalence of massively parallel multi-core processor architectures, adopting large DNN models in embedded systems remains challenging, as most embedded applications are designed with single-core processors in mind. This limits DNN adoption in embedded systems due to inefficient leveraging of model parallelization and workload partitioning. Prior solutions attempt to address these challenges using data and model parallelism. However, they lack in finding optimal DNN model partitions and distributing them efficiently to achieve improved performance. This paper proposes a DNN model parallelism framework to accelerate model training by finding the optimal number of model partitions and resource provisions. The proposed framework combines data and model parallelism techniques to optimize the parallel processing of DNNs for embedded applications. In addition, it implements the pipeline execution of the partitioned models and integrates a task controller to manage the computing resources. The experimental results for image object detection demonstrate the applicability of our proposed framework in estimating the latest execution time and reducing overall model training time by almost 44.87% compared to the baseline AlexNet convolutional neural network (CNN) model. • Parallel Computing. • Machine Learning. • DNN Model Partitioning. • Embedded Systems. • Embedded Software. [ABSTRACT FROM AUTHOR]

In this paper, we propose a reliability optimization technique for low earth orbit (LEO) satellite systems that operate in highly varying temperatures. In the proposed technique, we harden the target satellite on-board computer considering two common reliability threats, soft-error and hard-error, at the same time. Based on an existing fault-tolerant scheduling, we show how the reliability and lifetime of the system can be quantified with respect to a given error handling policy. Then, using these models, we perform extensive reliability analyses and optimizations of the hardening degree for different functional safety requirements. We implemented the proposed technique in a widely adopted real-time operating system (RTOS), Real-Time Executive for Multiprocessor Systems (RTEMS), working on a real satellite on-board computer system, GR712RC. The proposed technique enables efficient explorations of the trade-off between the soft-error reliability (failure rate), hard-error reliability (expected lifetime), and other resource utilization indicators, like power consumption or CPU utilization. In particular, the proposed technique could be successfully utilized to co-optimize the lifetime and fault-tolerance concerning the unique ambient temperature profile of LEO satellites.

As the behavior of multitasking embedded software is dependent on the underlying operating system(s), rigorous and efficient verification in this domain requires models of operating systems (OS) that enable OS-aware verification of application programs at reduced cost. However, the heterogeneity of the languages used for OS models and of the program source code makes it difficult to compose these seemingly independent components and thus requires translation of one language into another, causing various issues in verification. To alleviate this problem, we propose a hybrid approach that composes formal OS models with application programs in an interaction model. Based on typical OS-application interaction behavior, our interaction model is a composition framework that connects an OS model to application programs as long as they share the same Application Program Interface (API). It provides seamless composition of two heterogeneous software artifacts by formulating source code annotations based on control-flow analysis and by synchronizing state transitions over API function calls to regulate the context switching of multitasking programs. A prototype implementation of the interaction model was applied to eight benchmark programs of the Erika OS and a control program with real-scale complexity from the automotive domain. It was shown that the framework supports systematic and effective verification of multitasking embedded software, which has not been possible using code-level model checking.

Our energy scenario is nowadays shaped by progressive electrification of energy final use. In this context, electricity networks are seeing a growing multitude of distributed assets entering from the edges of the grid and acquiring new ICT capabilities that were limited before to a restricted number of major players. Particularly, assets like Photovoltaic Inverter (PvI), Electric Vehicle (EV) chargers, wind turbines controllers, programmable loads, storage systems, and other Distributed Energy Resources (DER) are now able to communicate through different technologies and make conscious choices under human-decisions or even independently. This is leading to a decentralization of the system's view by increasing single actor independence. Notwithstanding, a problem arises when current centrally-managed electricity networks struggle to coordinate massive amounts of new figures and adapt to this new decentralized paradigm. Therefore, a decentralized coordination-and-control framework will ensure better integration of s and new figures as prosumers, while allowing higher exploitation of their potential compared to centrally managed systems. This article seeks in Blockchains the enabling technology for designing and supporting such a grid infrastructure. It develops a first framework to address this need by envisioning a grid-system based on the direct participation of nowadays-used embedded-energy-devices within a decentralized platform hosting specific coordination procedures. The platform was developed in an experimental research campaign performed at ABB Laboratories basing on embedded-devices currently designed as control-connectivity boards for smart-inverters. Therefore this article introduces the background theory and reasons behind this proposed system. The intent here is not to give all the specific details of the implementation, but introduce the supporting reason, high-level design, and required characteristic of the Blockchain-based platform for coordinating grid operations. Blockchain technology is seen here as the appropriate technology to enable the realization of a multi-actor energy-management system and enable distributed coordination in power grids.

The advent of machine learning (ML) methods for the industry has opened new possibilities in the automotive domain, especially for Advanced Driver Assistance Systems (ADAS). These methods mainly focus on specific problems ranging from traffic sign and light recognition to pedestrian detection. In most cases, the computational resources and power budget found in ADAS systems are constrained while most machine learning methods are computationally intensive. The usual solution consists in adapting the ML models to comply with the memory and real-time (RT) requirements for inference. Some models are easily adapted to resource-constrained hardware, such as Support Vector Machines, while others, like Neural Networks, need more complex processes to fit into the desired hardware. The ADAS hardware (HW platforms) are diverse, from complex MPSoC CPUs down to classical MCUs, DPSs and application-specific FPGAs and ASICs or specific GPU platforms (such as the NVIDIA families Tegra or Jetson). Therefore, there is a tradeoff between the complexity of the ML model implemented and the selected platform that impacts the performance metrics: function results, energy consumption and speed (latency and throughput). In this paper, a survey in the form of systematic review is conducted to analyze the scope of the published research works that embed ML models into resource-constrained implementations for ADAS applications and what are the achievements regarding the ML performance, energy and speed trade-off.

We consider the problem of adapting embedded software to heterogeneous devices where it is impractical to obtain a system-level power model for each target platform and operating environment. Our solution leverages the emerging capability of measuring power consumption at run-time using a built-in battery monitoring unit (BMU). We use a statistically rigorous design of experiments (DoE) methodology to efficiently characterize the power consumption configuration space of software online instead of constructing a system-level power model offline. This approach is simple, low-cost, and permits software to dynamically select suitable parameters at run-time to satisfy energy or performance constraints. We illustrate this approach using a camera trap application as a case study deployed on a Nokia N80 smartphone. © 2014 IEEE.

LiDAR sensors are extensively used in autonomous vehicles and their optimal use is a critical concern. In this paper, we propose an embedded software architecture for multiple LiDAR sensors, i.e., a fusion system that acts as an embedded system for processing data from multiple LiDAR sensors. The fusion system software comprises multiple clients and a single server. The client and server are connected through inter-process communication. Multiple clients create processes to process the data from each LiDAR sensor via a multiprocessing method. Our approach involves a scheduling method for efficient multiprocessing. The server uses multithreading to optimize the internal functions. For internal communication within the fusion system, multiple clients and a single server are connected using the socket method. In sequential processing, the response time increases in proportion to the number of connected LiDAR sensors. By contrast, in the proposed software architecture, the response time decreases in inverse proportion to the number of LiDAR sensors. As LiDAR sensors become increasingly popular in the field of autonomous driving, the results of this study can be expected to make a substantial contribution to technology development in this domain.