Search

This paper introduces a novel model-driven methodology for the software development of real-time distributed vehicular embedded systems on single- and multi-core platforms. The proposed methodology discloses the opportunity of improving the cost-efficiency of the development process by providing automated support to identify viable design solutions with respect to selected non-functional requirements. To this end, it leverages the interplay of modeling languages for the vehicular domain whose integration is achieved by a suite of model transformations. An instantiation of the methodology is discussed for timing requirements, which are among the most critical ones for vehicular systems. To support the design of temporally correct systems, cooperation between EAST-ADL and the Rubus component model is opportunely built-up by means of model transformations, enabling timing-aware design and model-based timing analysis of the system. The applicability of the methodology is demonstrated as the proof of concepts on industrial use cases performed in cooperation with our industrial partners.

According to the model-driven engineering paradigm, one of the entry requirements when realizing a seamless tool chain for the development of software is the definition of metamodels, to regulate the specification of models, and model transformations, for automating manipulations of models. In this context, we present a metamodel definition for the Rubus component model, an industrial solution used for the development of vehicular embedded systems. The metamodel includes the definition of structural elements as well as elements for describing timing information. In order to show how, using model-driven engineering, the integration between different modeling levels can be automated, we present a model-to-model transformation between models conforming to EAST-ADL and models described by means of the Rubus component model. To validate our solution, we exploit a set of industrial automotive applications to show the applicability of both the Rubus component model metamodel and the model transformation.

The existing worst case response-time analysis for controller area network (CAN) with nodes implementing priority and First In First Out (FIFO) queues does not support mixed messages. It assumes that a message is queued for transmission either periodically or sporadically. However, a message can also be queued both periodically and sporadically using mixed transmission mode implemented by several higher level protocols for CAN that are used in the automotive industry. We extend the existing analysis for CAN to support any higher level protocol for CAN that uses periodic, sporadic, and mixed transmission of messages in the systems where some nodes implement priority queues, whereas others implement FIFO queues. In order to provide a proof of concept, we implement the extended analysis in a free tool, conduct an automotive-application case study, and perform comparative evaluation of the extended analysis with the existing analysis.

Convolutional neural networks (CNNs) provide the best accuracy for disparity estimation. However, CNNs are computationally expensive, making them unfavorable for resource-limited devices with real-time constraints. Recent advances in neural architectures search (NAS) promise opportunities in automated optimization for disparity estimation. However, the main challenge of the NAS methods is the significant amount of computing time to explore a vast search space [e.g., $1.6\times 10^{29}$ ] and costly training candidates. To reduce the NAS computational demand, many proxy-based NAS methods have been proposed. Despite their success, most of them are designed for comparatively small-scale learning tasks. In this article, we propose a fast NAS method, called FastStereoNet, to enable resource-aware NAS within an intractably large search space. FastStereoNet automatically searches for hardware-friendly CNN architectures based on late acceptance hill climbing (LAHC), followed by simulated annealing (SA). FastStereoNet also employs a fine-tuning with a transferred weights mechanism to improve the convergence of the search process. The collection of these ideas provides competitive results in terms of search time and strikes a balance between accuracy and efficiency. Compared to the state of the art, FastStereoNet provides $5.25\times $ reduction in search time and $44.4\times $ reduction in model size. These benefits are attained while yielding a comparable accuracy that enables seamless deployment of disparity estimation on resource-limited devices. Finally, FastStereoNet significantly improves the perception quality of disparity estimation deployed on field-programmable gate array and Intel Neural Compute Stick 2 accelerator in a significantly less onerous manner. [ABSTRACT FROM AUTHOR]

One of the challenges in Delay Tolerant Wireless Sensor Networks (DT-WSN), is to handle situations where the available buffer space is insufficient– the buffer management problem. Although several buffer management algorithms have been proposed for DT-WSNs, to the best of our knowledge, there is no comprehensive study on the effects different factors have on their performance, and which evaluates the relative performance of these algorithms in different contexts. This paper evaluates in a fixed-factor factorial experiment the performance in terms of latency and Quality of Information (QoI) of four representative buffer management algorithms for DT-WSNs; two traditional, FiFO and Random, and two QoI-based algorithms– one proposed by Humber and Ngai and the SmartGap algorithm. The evaluation suggests that the buffer management algorithm in combination with employed routing protocol and the sensor node buffer sizes have a significant impact on latency, while the obtained QoI rather depends on the characteristics of the transported data and the routing protocol, provided a single-copy routing protocol is used. Moreover, the evaluation suggests that QoI-based buffer management algorithms do offer improved QoI, with an 31% improvement in MAE for SmartGap compared to FIFO. However, they do so at the expense of higher latency, with SmartGap giving a 60% higher latency than FIFO on average. [ABSTRACT FROM AUTHOR]