Multischedule Synthesis for Variant Management in Automotive Time-Triggered Systems

Abstract

Car manufacturers provide a growing variety of models and configuration options for customers. In the highly competitive and cost-driven automotive industry, managing these variants and increasing the reuse of functionality in different variants has therefore become one of the key challenges. This paper addresses the problem of generating variant schedules for time-triggered Electrical/Electronic (E/E)-architectures. We propose a multi-schedule synthesis approach that determines the common parts of multiple variants and generates a schedule that exploits this commonality. Hence, a multi-schedule defines individual variant schedules with an identical schedule for applications common to different variants. This makes these applications variant-independent, thus, reduces the testing and integration efforts as it only has to be done once. Multi-schedule synthesis involves several challenges, viz., identification of commonality between different variants, schedule synthesis for common parts, and the integration of uncommon parts. Consequently, the schedule synthesis approach presented here is very different from conventional approaches. Finally, to address the increased complexity, we also propose a divide-and-conquer approach to partition the problem, improving the scalability.

Multi-Variant Scheduling of Critical Time-Triggered Communication in Incremental Development Process: Application to FlexRay

“…Moreover, these vehicle models also have many versions (e.g., a configuration with an adaptive LED frontlight system or with common halogen lamps, etc.). Thus, an efficient variant management can be a significant economical and competitive factor [4].…”

“…A proposal for Multi-variant Scheduling was first presented in [23], wherein schedules for more variants were created, all at once. This was later followed up by Sagstetter et al [4] who have iteratively constructed a multi-schedule, in which the signals common to all the variants are scheduled in the first iteration, the shared signals in the intermediate iterations, and the signals specific to just one variant in the last iteration. To the best of the authors' knowledge, all the published methods for scheduling the time-triggered communication have been built on greenfield, without considering previous/original schedules.…”

“…Variants of the software product and their source codes need to be managed carefully [8], [9] otherwise the product becomes uncontrollable with smelly code [10]. Sagstetter et al [4] observe that, by rapidly increasing number of vehicle model variants, variant management becomes an important and challenging problem for the fast evolving automotive industry. They have shown that vehicle variant management is strongly linked to the creation of time-triggered communication schedules.…”

Semantics-Preserving Cosynthesis of Cyber-Physical Systems

“…Here, the design is mostly aimed at task mapping, frame packing, and task and message scheduling [32]- [35]. Several works also consider the combined synthesis of task and message schedules from the application-level timing requirements such as latencies and response times [36]- [39]. In these approaches, usually a constraint programming (CP) problem is formulated and solved with integer linear programming (ILP) or satisiability modulo theories (SMTs) solvers, heuristics or metaheuristics methods.…”

Architecture exploration for distributed embedded systems: a gap analysis in automotive domain

“…The number of manifestations of one ECU should also be minimized during the optimization to increase the utilization of that ECU [19]. The schedule of tasks on ECUs with multiple manifestations can be designed by the method introduced in [38]. In addition, when optimizing the whole product line, estimated sales volumes of different variants should also be considered together with the uncertainty of the estimation [20].…”