IVaM: Implicit variant modeling and management for automotive embedded systems

Abstract

In this paper, we propose a graph-based approach for the modeling and efficient analysis of functional variants of a car's electric and electronic (E/E) architecture functionality by combining local technical expert knowledge with global business knowledge. Starting with a variants system specification including a set of task graphs, linear constraints on binary variables are specified for their alternative selection as well as the selection of groups of alternatives called application groups. These constraints may stem from a certain domain knowledge, e. g., entertainment or power train domain, or global constraints. The typically vast space of resulting possible combinations of different selections of alternative behaviors will be termed variant space and those satisfying the set of formulated constraints valid variants. An important result of this paper is that the set of variants, and especially the set of valid variants, do not need to be modeled or stored explicitly but rather implicitly. Nevertheless do we show that using state-of-the-art PB solver techniques, we may determine the set of valid variants very efficiently. Each of these valid variants may subsequently be used as a candidate for design space exploration (DSE) in order to optimize also the mapping of the corresponding task graph functionalities to a final optimized E/E architecture. A real-world case study is provided to demonstrate the capabilities and efficiency of the presented approach on implicit variant modeling and analysis.

Multifaceted Automated Analyses for Variability-Intensive Embedded Systems

“…Other approaches [57]- [59] tackle both functional and platform variabilities by focusing on an optimal platform configuration for a multi-variant application. Contrarily to our solution, they do not find a design (i.e., a mapping of a functional variant onto a platform configuration) that ensures an optimal execution.…”

“…Intuitively, these graphs encode two forms of variability. The first consists of variations in the data flow, as we allow data paths to have multiple connected producers and consumers; this follows the same approach as in variable workflows [57]. The second lies in the alternative attribute values that a node can take.…”

Functional feasibility analysis of variability-intensive data flow-oriented applications over highly-configurable platforms

“…al. [13,14] manages some variability in both platform and functional specifications. On the platform variability side, resource components can be selected or not, while optional and mutually exclusive task groups are managed on the functional part.…”

“…To implicitly represent all these variants in a single model, we follow the same approach as in variable workflows from [13], allowing data-paths to have multiple input and output tasks connected.…”

Verification and validation of embedded software in an automotive context: a systematic literature review

“…Embedded automotive software is usually a low-cost solution that can be quickly implemented compared with hardware development (KRIEBEL et al, 2018). An automotive vehicle is a complex system with many components such as sensors, actuators, and the Electronic Control Unit (ECU), and the hardware components are usually managed by embedded software (GRAF et al, 2013). The increasing complexity of software-controlled systems has driven discussion on the robustness of current frameworks for testing software, especially owing to issues with autonomous cars.…”